True to pattern, Paul Levy of Beth Israel Deaconess thinks it's a good idea for hospitals to make their HSMR scores public, and wonders why insurers don't require that. He frames it as a matter of keeping good faith with the public. Now, his protestations notwithstanding, that's difficult to disentangle from the fact that BIDMC has always performed quite well on HSMR — considerably below the national average. Now, you'd think that with all of our totally fabulous health care in Massachusetts, we'd be doing pretty well, right? Well, no: Massachusetts hospitals as a whole have tracked about ten points worse than the national rate since the 2000 data. And there are a few hospitals in MA that continue to be far, far worse in this metric than the national average (i.e. 140, vs. 100). At the very least, should they not know internally that this is the case? As BIDMC's Ken Sands posits, knowing the HSMR is useful for generating hypotheses within the hospital about how to improve care.
Followers of the standardized-testing debate in schools will recognize this pattern: Some people don't like being evaluated on the basis of a single numerical score, like Mass. General CEO Gregg Meyer. (To his credit, Meyer was a good enough sport to show up and make his case.) Meyer notes that “transparency” is the “mom and apple pie” virtue of business these days, but worries about popular misinterpretation of medical statistics. And indeed, much like their college rankings, hospital “rankings” of the US News and World Report kind are probably worth a bucket of warm spit.
But it strikes me that the answer is better statistics — and better, more sophisticated public interpretation — not to bury them altogether. Meyer himself mentioned a quality-improvement effort in Pittsburgh, wherein the local media were pro-actively briefed on how to interpret data. Such trust-building efforts would be most welcome — especially as opposed to silence.
We commonly see statistics used to bewilder and obfuscate rather than enlighten: For instance, baseball fans have long put too much emphasis on a batter's batting average, at the expense of say, on-base percentage. That being said, is batting average completely irrelevant? Moreover, should it be hidden from the public? If a hospital has higher mortality rate than it ought, that should be public; otherwise, we're going on the “trust-me” model of public accountability. The age of revered, unquestioned institutions — the courts, medicine, churches —
is over. Particularly in an era of increased public involvement and dollar-investment in our health care system, it's not too much to ask that our medical providers earn their credibility.
There are some basic problems with using mortality rate as a measure. Consider these situations:
<
p>
1) Suppose Hospital A has an international reputation for cancer care. All of the truly desperate cases will come from around the world to get their care. Being desperate cases, they’ll often die getting that care. Does that mean Hospital A is worse than the county hospital down the street that only gets “normal” cases and therefore saves them at a higher rate?
<
p>
2) Population demographics — a hospital with a very old or poor population will have a higher mortality rate than one with a young, affluent population, right?
<
p>
The stats shouldn’t be hidden from the public (I can’t think of very much data that should be hidden from the public, to be honest), but they need to be released with a whole lot of caveats. The raw numbers won’t help people choose a good hospital — they need explanation, too.
that HSMR actually does calculate those factors — or at least it can.
<
p>
It was agreed by all concerned that there is only so much value for this as a comparative consumer tool — should I go to hospital A or B — but it is useful to the hospital, at least to know whether care is getting better or worse.
The HSMR score represents the ratio of actual deaths to expected deaths, were the latter takes into account diagnosis and demographic factors.
That makes a lot more sense. I’m sorry if that was somewhere in the links and I missed it — I did look around before posting, but didn’t see that.
<
p>
Doesn’t change my ultimate opinion that the information should be available to all concerned (i.e., the public, too) along with clarifying warnings and information.
<
p>
Thanks for clearing up my misconceptions.
I don’t know if this will make sense to you, but Sir Brian Jarman’s PowerPoint from the event is here. In it he explains a little bit how the metric is put together.
I agree: the answer to incomplete information is more information, in every public sector. Paul Levy is a hero!
<
p>
Status quo protectors — ie, people who work in the sector, who usually are well-financed and tend to contribute heavily to politics — try to prevent us (public) from getting add’l information by setting up an impossible-to-reach standard that says “Unless all information is somehow perfect, none can be released.”
<
p>
For example, there was a Globe story 2 weeks ago that the cops weren’t providing requested info on the race of drivers they stopped. Ah, here it is.
<
p>
<
p>
Why?
<
p>
<
p>
I don’t do a lot of citizen traffic stops, but I’m not clearn on how checking off race and gender can take more than a couple seconds.
<
p>
And even if the chiefs were RIGHT, the data could serve a valuable purpose, which is to assauge a significant chunk of citizens who think otherwise.
<
p>
Data should need to pass the “reasonable-ness” test, not the “perfection” test, in order to be shared. And the gatekeepers shouldn’t be the self-interested.
<
p>