05 de març 2015

Practice makes perfect

Comparing hospital performance within and across countries: an illustrative study of coronary artery bypass graft surgery in England and Spain

My concern over variations in clinical practice relies on a specific issue. Once you've describe it, you need to understand its implications. Thus, somebody should assess whether variations cause poor health outcomes. Before starting such a task, somebody has to measure relative performance, and this is precisely what a recent article in EJPH does on CABG surgery in England and Spain. I would like to highlight this statement:
In this article, we use patient-level data within and between two countries to assess the added value of pooling administrative data across countries and to explore hypotheses that may explain differences such as those reported in cardiac care. These may be driven by a small number of hospitals with unacceptably high mortality rates (perhaps due to coding differences or under-performance). Otherwise, country differences in outcomes may be explained by the concentration of services into specialist centres with differences in clinical facilities and staff experience, as reflected by hospital volume of surgery. These hypotheses cannot be tested adequately using within-country data or national aggregates, but lessons may potentially be learned from hospital-level comparisons across countries using comprehensive administrative data.
If we focus on performance, national aggregates confound. And this is focus of the article:
Unadjusted mortality rate following CABG surgery demonstrates a considerable difference between hospitals (particularly in Spain) and between countries (average mortality is 2.3% in England, 5.0% in Spain)
After adjusnting and pooling data from both countries, then results look different:
First, the hospitals’ performance contrasts substantially with the traditional within-country findings. Nine Spanish hospitals are identified as ‘alarms’ in the pooled assessment compared with five in the country-specific assessment. Thirteen Spanish hospitals are additionally identified as ‘alerts’ that were within the normal range when considering Spain alone. Four English hospitals are now identified as alerts and none is assigned alarm status. Second, there is a clear separation in the number of expected deaths between English and Spanish hospitals, reflecting differences in volume across countries. The median hospital surgical volume in Spain is 154 patients a year, compared with 690 in England, and the highest volume hospital in Spain treated 337 patients in 1 year, whereas the lowest volume hospital in England treated 327. Third, despite the large overall between-country difference, the vast majority of hospitals in England and around a third of those in Spain lie within or below the 95% funnel and are largely comparable in terms of their SMR.
This is an excellent explanation of "practice makes perfect" argument. And, if this were the only factor, there is a compelling reason to concentrate CABG surgery in certain hospitals and close services in others. We know that some concrete hospitals may have high adjusted mortality rates and deserve a concrete action. Urgent decision is needed, just to reduce mortality ratio by half.

PS. The whole issue on variations in EJPH represents a milestone in health services research. Congratulations to the authors and the ECHO project.

PS. GCS blog on the same topic.

PS. New book available: The Triple Aim for the future of health care by Núria Mas and Wendy Wisbaum