Start with benchmarking
While disease management (DM) programming intuitively makes sense to many occupational health professionals, it has been difficult to demonstrate that such programs actually save money.
One of the impediments in the past has involved less than ideal benchmarking techniques. What statisticians call regression to the mean can result when participants are identified with a condition during a high point of medical utilization, when treatment is most costly. Because of the inherent bias of such a setup, probabilities can favor a return to average utilization even without any DM intervention.
Richmond, VA-based Health Management Corporation (HMC) appears to have overcome some of these obstacles with the use of a control group that was statistically similar to the group receiving the DM intervention, but was not offered the program services. This, HMC statisticians claim, eliminates the effects of selection bias because the control group did not opt out of the program.
In a unique position
Since HMC was owned by a managed care organization, namely Anthem Southeast, it was able to obtain data for a control group as well as for the study group. The control group was drawn from Virginia employees who were not eligible for HMC’s DM services, because the benefit managers in the self-insured PAR/PPO (Participant/Preferred Provider Organization) did not choose to use those services for their workers. The study group was composed of employees eligible for the DM services.
"We were in a unique position, so it was advantageous for us to do the study this way," notes Michael Cousins, PhD, a biostatistician who is HMC’s director of informatics.
There were 76,194 members and 2,359 diagnosed participants in the study group. Each group was tracked for the year prior to DM implementation (1999) and the year following implementation (2000). The study covered most, but not all, of HMC’s DM offerings. "We looked at asthma, coronary artery disease, congestive heart failure and diabetes," Cousins reports. "In this study, we did not analyze congestive heart failure because there were only 24 people who qualified."
Study yields positive results
The study showed that the DM programs did, in fact, save money. Preliminary analyses showed that they produced net savings of 94 cents, per member per month, for total net savings (after the cost of the program) of $859,471. The term "net program savings" means the actual costs for all the people in the study group were $859,471 less than what could be expected for the control group. To get the 94 cents the researchers divided that by total member months.
The study group’s medical claims costs to the plan were $12,048,275. Gross and net savings as a percent of the study group’s claims were 11% and 7.1%, respectively. Claims costs for the 76,194 members studied totaled $124,029,000, and net savings as a percent of claims for the whole population were 0.7%.
"We subtracted the cost of the program [to reach the net figure]," notes Cousins. "So, for the people in study, we had total claims costs of just over $12 million. We took the $859,471 and divided it by 12 million to get the 7.1%. That is the proof of the pudding."
In addition, he notes, all dollars were normalized. "We had actuaries as well as statisticians working on the study," he explains. "Normalized means we took the costs for procedures in both groups and calculated the average cost for those procedures, and applied it to both groups. This way, we ensured that the hospitals contracts for the control group weren’t biased vs. those of the study group."
Furthermore, Cousins and his colleagues employed statistical simulations that showed there is an 82% chance there will always be savings with this program.
Being a biostatistician, he is conservative by nature. Nevertheless, he is comfortable with the conclusion that the program saved money. "Another good thing I can say is the methodology we used avoided selection bias and regression to the mean, so I certainly have greater confidence in the results," he observes. "There are not a whole lot of companies out there in the DM industry who use this rigorous methodology."
[For more information, contact:
• Health Management Corp., 6800 Paragon Place, Suite 500, Richmond, VA 23230. Telephone: (800) 523-9279. Fax: (804) 354-5047. E-mail: firstname.lastname@example.org.]