Monday, April 21, 2008

Health Dialog takes the World Health Care Congress on a Deep Dive and We Liked It

A thoroughly caffeinated and excited-to-learn Disease Management Care Blog attended the first day of the World Health Care Congress. While there were a host of sessions presided over by the lesser and greater gods of the health care cosmos, the most interesting was hosted by Susan Dentzer (soon-to-be Editor in Chief, Health Affairs) on the “Deep Dive” clinical trial. The two presenters were David Wennberg MD, MPH of Health Dialog and Lance Lang MD of Health Net, Inc. Almost 190,000 (not a mistake, that’s 190 thousand) commercially insured patients from Highmark Blue Cross Blue Shield (east coast) Health Net (west coast) were randomly assigned to 'usual' care management (serving as the control group) versus “enhanced” care management (the intervention group).

What difference in between the two groups? According to Dr. Wennberg, the intevention group was the subject of 'analytics on steroids.' As the DMCB understands it, this consisted of a later-generation predictive modeling capability that was hypothesized to do a better job of identifying which patients would benefit from Health Dialog’s remote patient coaching.

As many in the population-care ‘business’ already intuitively know, not all patients with a chronic condition necessarily benefit from disease management. For example, many patients implacably prefer not to be called, many have already achieved maximum self-care, others are not ready to make any life-style changes, many are victims of other random illness, many have severe disease that may not benefit from remote coaching and many have other concurrent conditions that are not amenable to any intervention. The DMCB suspects that Heath Dialog used its prior experience in the art and science of patient engagement to develop a predictive modeling tool that culled subjects who would be most a) open to and b) likely to benefit from its care programs. By targeting the “health coaching” at the ‘optimum’ patients in the intervention group, the researchers suspected total health care costs would be lower versus a control group that received the relatively untargeted health coaching.

The content of the health coaching was not different in the two groups; what was different is that the Health Dialog nurses received different lists of patients to call. The patient randomization occurred at a household level. Following randomization, the control and intervention groups were statistically similar by mean age, male-female ratio, disease burden and baseline costs. In classic Health Dialog style, the two groups were also similar according to the proportion of persons with the a) ‘big 5’ diseases, b) ‘preference sensitive conditions’ (examples include heart disease amenable to surgery, hip and knee arthritis, back problems and uterine disorders) and c) ‘coachable’ conditions (examples include obesity, hyperlipidemia, migraine and abdominal pain).

Bottom line: compared to the control group, costs were significantly lower in the intervention group. The majority of savings appeared to be due to lower inpatient utilization, but this was also true across the board including emergency room visits and outpatient visits. The savings weren’t huge, but enough to cover the cost of the program and then some. What’s more, Lance Lang presented evidence from Health Net showing a compelling impact on overall trend. As a result of these data, Health Net and Highmark have moved their control patients into the ‘analytics on steroids.’

While the DMCB and its readers will need to await release of the detailed information in a public peer review setting, this preview is interesting at several levels. First of all, Health Dialog and its partners are showing “applied” health services research is possible in the business setting. Secondly, while critics of disease management fault the industry for failing to show any 'return on investment,” that criticism is really being directed at Ver. 1 programs that are long gone. The industry has already moved beyond those early approaches with new programs. Third, modern strategies for persons with chronic illness may need to rely more on ‘market segmentation’ than on finding new remote engagement strategies. Fourth, there is no good news here on how this might work in Medicare population: this information doesn’t appear to be generalizable to that group.

Last but not least, Drs. Wennberg and Lang noted the success in this trial is not the “magic bullet,” but is an important consideration in a broader multi-pronged strategy aimed at controlling health care costs. I am pleased to report that Ms. Dentzer and the audience did not appear to disagree.

No comments: