Need a diabetes program? Database can help prove it

Sachs Group offers statistics for your zip code

You might have a sinking feeling that a particular disease is costing you more money than necessary. And you know tighter management of high-risk patients would probably improve their lives and save you money. But can you prove it?

Hard numbers on utilization differences among patients would go a long way toward justifying the cost of implementing a disease management program, but those numbers aren't always easy to come by. And even if you look at national numbers, they won't necessarily hold true for your patient population. Now there's a new option from the Evanston, IL-based consulting firm Sachs Group: a database that contains extensive prevalence and utilization statistics right down to the county and zip code level.

Sachs began with diabetes for its new Health Profiles series that studies utilization across the continuum of care and breaks the data down not only geographically but by age, sex, and high- and low-risk patients. The information was gleaned through a search of every available database, including state-by-state inpatient data, Medicare files, the National Health Interview Survey, and Sachs Group's own proprietary claims databases. Data on congestive heart failure and cancer should be released by the end of the summer.

"We wanted to provide people with empirical benchmarks rather than theoretical benchmarks," says Dennis Dunn, senior scientist for the Sachs Group. "That's important especially for diabetes, which varies tremendously from one county to another in severity, utilization, and practice patterns. Our conclusion has been that if you obtain information based on some other area of the country, the forecast is really unlikely to represent what is happening in your population."

Take New Orleans, for example. According to Sachs' Diabetes Health Profile, the local prevalence of diabetes is 4.3% across all age groups, compared with 3.5% in the South and 3.4% nationally. "That's at least 25% higher than the rest of the country. If you made a forecast based on national information or even on Southern regional information, you would have a severe underestimate of your prevalence," Dunn says. "If you planned a program counting on 3.4% prevalence, you could be dreadfully underestimating the utilization of resources to treat that population."

The database also shows what the effect would be of managing the high-risk group - the 15% to 20% of diabetics who use resources disproportionately - down to the utilization of the rest of the population, Dunn says. So the rest of the local area becomes the benchmark for the sickest diabetics. "Often you'll see benchmarks for utilization decreases for well-managed diabetics that we feel are unrealistic, that aren't based on data but on normative standards for what patient days for diabetics ought to be," Dunn says. In New Orleans, a potential decrease in hospital days of 15% to 16% for all diabetics could be expected.

If you look at national statistics for skilled nursing days, for example, you'll get an average that likely won't apply to your area. That's because utilization of skilled nursing days is low in some areas simply because the facilities aren't available, Dunn says. Another potential pitfall is looking at clinical trial information that shows reductions based on closely managing a select group of patients, not an entire population. "Those forecasts assume that you have the resources to follow and tightly manage each individual over a long period of time, which is pretty unrealistic," he says. "Diabetes treatment programs have to deal with noncompliance, with patients dropping out, with not being able to contact large segments of their population at all."

Once you have local statistics in hand, you can use them to help you develop a diabetes treatment program, Dunn says. If you're going to assume some or all of the risk on your program, you'll need realistic expectations of patient days, emergency room visits, and other variables to help you set goals and rates. And you need a realistic idea of what your enrollment will be down the road. You also could use the information to decide who to benchmark against so you're comparing yourself against an area with similar prevalence and utilization rates.

[For more on the Diabetes Health Profile, contact Sachs Group, 1800 Sherman Ave., Evanston, IL 60201. Telephone: (847) 475-7526.]