Do You Speak English? Or: why can't I understand the government's statistics?

Don Berwick has an interesting interview with a good point:

I have come to think that transparency is one of the most important aspects of policy in the nation. I don’t think we’re where we need to be and I think if we can make more information more transparent, almost everything will get better, cost and price. I have become more and more a fan of transparency. In the Affordable Care Act, there are more and more tools for this. I think transparency, by which I mean the ready availability of data while protecting patient privacy, at low costs helps everyone. It helps providers compare their quality to one another. It helps buyers look at pricing. It helps consumers know more about the care they get or could get. Some of the consumers that want that information will make good use of it, but many will not. They’re too busy with other things. But overall the effect of transparency will be entirely favorable. Do I think patients will open Consumer Reports to decide where to go when they have a heart attack? Of course not.

It’s echoed in the opening post of US News & World Report’s “Second Opinions” blog—the idea that we just don’t have good enough data, particularly for underreported adverse events.

Fortunately we’re starting to get more data—for example, in California (“For the first time, Californians can compare infection rates at their hospitals in reports released this month by the state Department of Public Health.”) to say nothing of CMS’s “Hospital Compare” website (NPR’s Jordan Rau does some good work with infection rates from catheter infections here).

The trouble, as Berwick notes, is “so what?” Having all this data is useless unless it’s actionable—and who’s going to be dealing with the data? Berwick doesn’t think patients will open Consumer Reports to decide where to go when they have a heart attack, which I suppose opens up two related questions: if a hospital is demonstrably bad at treating heart attacks, and the heart attacked doesn’t have time to evaluate hospitals, who’s going to do it for him? (His insurer? Certainly that’s the way to read their move into changing the payment structure) The other question is, shouldn’t patients be comparing the hospitals via this data?

Obviously the first problem there is patient choice—there’s often not a lot of choice in the matter as to where and with whom you’re treated, for patients. The second problem is that a lot of these presentations of data are pretty useless. How many people know of Hospital Compare, or where to find rates of infection in hospitals in California? Low, of course. And then there’s the presentation of data.

Let’s take my local hospital, Strong Memorial. There’s a lot of data here. Unfortunately, there’s a lot of data here. There’s little attempt to contextualize it—how important is this stuff I’m seeing? What’s the most important data? There just tables upon tables—a wonk’s dream, but not a harried consumer’s. That’s when they provide data. In many categories, they don’t. If you want to compare Strong’s rate of deaths among patients serious treatable complications after surgery to the national rate, we’re only informed that it’s worse than the national average of 115.7 per 1,000 discharges. It surely makes a huge difference if it’s 116 versus, say, 400. (It works in both directions—Strong is better than the U.S. benchmark on central line blood stream infections, but we’re not told how much better (or, indeed, what the U.S. benchmark is anyway).

We’ve seen a move from various governmental bodies to translate complex documents into comprehensible forms—Elizabeth Warren; the move to require agencies to offer “plain English” rules; Kathleen Sebelius’s effort to ban fine print in insurance forms—and that effort could probably profitably be done for the government’s own statistics.