To the Editor.
—The key finding recently reported by Dr Miller and colleagues1 and featured as an Original Contribution was previously described.2 However, the previous interpretation was very different.The significant inverse relationship between likelihood of hospitalization and all-cause, 30-day, postadmission-modeled mortality (the hospital sample standardized mortality ratio for 30 days [HSMR30]) suggested to me that states with indications of easier access to care (higher likelihood of admission) were associated with lower-than-expected mortality rates. That inverse pattern also held for disease-specific admissions for congestive heart failure, pneumonia, and stroke.2 These findings were interpreted by me as evidence that hospitalization may actually confer survival benefit, in contrast to the underlying assumption of most small-area utilization analyses that hospitalization necessarily puts people at risk.The interpretation of Miller et al of this finding, that it is evidence of modeling bias, is argued primarily through use of econometric theory. Secondarily,
Huff ED. Measuring Quality of Hospital Care. JAMA. 1995;273(1):21-22. doi:10.1001/jama.1995.03520250035015