Steven I. Weiss (aka Canonist) is on the case. Here's a comparison he makes:
On this issue, if we take the number of metzitzah b'feh procedures per year of 2,500, and assume one infection for those infants per year (which for whatever reason seems to be a number that everyone feels comfortable tossing around), we can compare that 1/2,500 ratio to that of the population generally. Of the roughly 120,000 New York City births per year, the Department of Health says there are roughly 30 infections, or 1/4000. So if all those numbers prove out — which for infections generally, infections in metzitzah b'feh cases, and number of procedures per year have a lot of valence — there's a difference in likelihood that could be somewhere between 33% and 50%.
This is a pretty good first try at a non-technical comparison of the incidence of neonatal herpes in the MBP population and New York as a whole (disregaring Canonist's terminology, which isn't what I'd use). But for an accurate, epidemiologically valid comparison, beyond nailing down the numbers as specifically as possible (Agudath Israel has an interest in exaggerating the frequency of metzitzah, for example), one would have to find a control group which is as comparable as possible to the MBP population. Ideally, that would be an ultra-Orthodox group which doesn't perform MBP. I haven't looked for differential neonatal herpes rates by religion or ethnicity, but I doubt those are followed up on by the Department of Health. A more likely tack is to find a zip code heavily populated by non-MBP Charedim and find the incidence of neonatal herpes within that area.
Taking Weiss's comparison as the best we've got, there's one thing that must be said upfront: Neonatal herpes is underreported. The question is whether the underreporting is differential, i.e., whether MBP parents (or their pediatricians) underreport neonatal herpes more than, less than, or with the same frequency as non-MBP parents (i.e. parents city-wide excluding those performing MBP). I have no idea what the answer is, but engaging in complete speculation, I would imagine that (a) the MBP population, like most Charedim, are "plugged in" to perinatal care to an extent greater than their socioeconomic status would predict, and thus (b) they would tend to report cases of neonatal herpes greater than other parents in the city, by and large, so that (c) underreporting would be greater among New York parents as a whole in comparison to MBP parents. This would overestimate any reported difference in incidence between MBP and non-MBP births.
I think there would be some interesting research here. More important than that, though, is what the DOH is doing to encourage reporting of neonatal herpes and the minimization of MBP.