Last week’s New England Journal of Medicine had a real wake-up call in it. The article “Variation In Hospital Mortality Associated With Inpatient Surgery” (Ghaferi et al 2009) had some stunning revelations. They demonstrated, using data from the National Surgical Care Improvement Project (SCIP), that the two-fold variation in surgical mortality rates amongst hospitals is not explained by the characteristics of the patients or by the occurrence of complications. Complication rates, in fact, were quite similar at high-mortality hospitals and low-mortality hospitals. What differed, however, were the mortality rates in those cases where major complications occurred. The study thus lends credence to the concept raised by Silber et al (Silber 1992) of “failure to rescue” as the major explanation for differences in mortality across hospitals. Essentially what it implies is that the variation in mortality rates is due to differences in the way hospitals react to and manage complications. Though the current study just looked at surgical mortality rates, we’d bet the findings would be similar if a number of medical conditions were studied.
We have long looked at some of the public “quality” report cards on hospitals and been skeptical about inferring anything about quality based on the mortality rates or complication rates in those databases. They have largely been based on administrative data (submitted for reimbursement purposes) rather than clinical data. The severity adjustments in those have been questionable. And when you see hospitals with low complication rates but long average lengths of stay, you can be pretty sure the hospital is probably underreporting complications. But the new study by Ghaferi et al. is different. It did not rely on administrative data. Rather it used the clinical data from the SCIP project that was well risk-adjusted, clinically detailed, and identified complications accurately.
So does this mean we should no longer spend our efforts trying to prevent complications? Of course not. Patient safety efforts to prevent complications are still critical from both quality and cost-effectiveness standpoints. But it does tell us we need to re-examine how we identify complications and how we respond to them. Ghaferi et al. speculate that studies showing improved quality with better nurse:patient ratios may be one of those factors. Staffing and processes in the ICU may be another area of difference. This may also stimulate further research efforts into computer surveillance programs that, for example, can identify sepsis earlier than even the most skilled clinicians can.
What is clear is that we are going to hear a lot more about this phenomenon over the next year.
Ghaferi AA, Birkmeyer JD, DimickJB. Variation in Hospital Mortality Associated with Inpatient Surgery. The New England Journal of Medicine 2009; 361:1368-1375
Silber JH, Williams SV, Krakauer H, Schwartz JS. Hospital and patient characteristics associated with death after surgery: a study of adverse occurrence and failure to rescue. Med Care 1992; 30: 615-29