In her response to our defense of Farr and Nightingale [1], Dr. Iezzoni fails to distinguish between two issues: the correct calculation of the incidence rate and the comparability of data across hospitals. In her original contribution, Dr. Iezzoni had accused Farr and Nightingale of using the wrong method of calculation and thereby making the wrong comparison [2].

We are amazed by the first part of her counterexample, the myocardial infarction rate of 326%. We gave exactly this example in our letter and at the same time indicated the solution: a 326% death rate per patient-year amounts to 0.9 deaths per 100 patient-days, which is not confusing.

Next comes the comparability of incidence rates between hospitals. In 1948, Greenwood stated that even if one uses the right denominator (as Farr and Nightingale did), comparisons between hospital statistics remain hazardous [3]. We have discussed this problem elsewhere [4]. One factor that Dr. Iezzoni mentions is a difference in length of stay between hospitals. When length of follow-up is related to mortality, which it almost always is, the simple solution is to stratify by time. That is done routinely, for example, by Cox proportional-hazards models that estimate hazard rates (which are ratios of incidence rates by the day!). When person-time denominators are used over longer periods of time, follow-up is subdivided into "first week," "second week," and so on. This adjustment for length of stay has nothing to do with the calculation of incidence. Of course, the solution proposed by Dr. Iezzoni-to use a fixed time interval-is an excellent one. It permits one to calculate either incidence rates or cumulative incidences afterward.

Did Farr and Nightingale have political motives? Of course they did; they bore a grudge against certain hospitals with poor ventilation and sanitation [5]. In this, they resemble Dr. Iezzoni, who militates (rightly) against the simplistic use of between-hospital comparisons [5]. Might Farr and Nightingale's zeal have led them to throw caution to the wind when publishing comparisons that proved how right they were? Not unlikely. In the end, were they right by insisting on reforms? Most probably. That, and not the way in which they did their calculations, is the heart of the argument.