Thursday, October 16, 2008

Health Ratings

RateMyHospital.ie was mentioned on this blog before. It is an Irish site that intends to produce a 'snapshot' of patient opinion at a given time on the level of hospital service that patients were offered. An American firm, HealthGrades, takes a different approach. Medical centers get one, three or five stars based on how many patients develop complications and die after receiving treatment. One star represents a poor performance. Three stars stands for "as expected." Five stars goes to the very best performers. Read more in the Chicago Tribune here. Some interesting details below.

The rankings are based on three years of data from Medicare and are adjusted to account for how sick the hospital's patients are... Consumers can look up the HealthGrade rankings by state and by condition.... Overall, HealthGrades says patients have a 70 percent lower chance of dying in a five-star hospital compared with a hospital with a one-star ranking.

2 comments:

Liam Delaney said...

havent looked at this fully but surely you would have to control for the characteristics at intake. Hospitals in low SES areas treating patients with multiple other conditions on intake are likely to score lower on the measure you describe Martin even if they were perfect hospitals. Similar argument we made a while ago about using MBA starting salaries as the measure of the quality of the MBA.

Martin Ryan said...

Thats a fair point Liam. Adjustments are made by HealthGrades to account for how sick the hospital's patients are. It could be argued that this controls somewhat for hospitals in low SES areas having a different kind of intake.

But I think the unobserved counterfactual is what you are essentially getting at. If any
(randomly assigned) patient was moved from a one-star hospital to a five-star hospital, would they have a 70 percent higher chance of surviving? The answer is most probably no. But would we expect some improvement in survival probability and longevity? If this were the case, any difference in percentage chances of survival could be a more accurate measure of hospital quality. Of course, we might also find out that the one-star could be better.

There's an analogy here with the debates around school league-tables; particularly that some schools serve a much different cohort of students than others.