hospital birth neonatal mortality given that the planning status of hospital births is unknown. For example, there are no data on the number or percentage of women who begin laboring at home but are transferred to hospitals because of complications. Compounding the challenge is variation in risk. Vital data do not provide enough information about risk. MacDorman expressed reluctance to analyze mortality data given the apples-to-oranges comparison involved when planning status and risk are unknown.
In response, Paneth stressed that vital data are only a starting point, but said, “I would not go so far as to say that because we cannot really fully answer, you should not answer at all.” He suggested that analyzing actual deliveries provides at least a sense of what those rates are in the different settings and whether there are any unusual mortalities. While it may be difficult to select comparison groups for studies (e.g., women in different settings but with similar risk profiles), vital data on U.S. birth certificates nonetheless provide enough descriptive information such that unusual events that warrant further investigation, what Paneth calls “sentinel events,” should stand out. He said, “Even without a denominator, their existence is of interest.”