Roberto Muehlenkamp has asserted (without any serious argument) that 3rd percentile female weights from the CDC data, which I’ve used to highlight the unreasonableness of his assumptions concerning weight, are too high to represent the mean weight of a poorly fed population such as the Polish Jews. Fortunately, we have a source of data that allows us to test this claim, namely a paper on children born in the Leningrad State Pediatric Institute during 1942. During the siege of Leningrad – a topic on which Muehlenkamp has expressed himself forcefully in the past – starvation conditions prevailed. The weights of children born during 1942 were recorded (at least in certain hospitals) and compared to the figures for previous years.
Body weights at birth did, naturally, decrease during the siege. Here are figures on the numbers of births:
Here are figures on birth weights:
(Note: while this data falls under the section header “Average Birth Weight of Children Born in 1942”, the analysis that follows it – e.g. giving 49.1% as the proportion of births under 2500 g during the first half of 1942 – indicates that these figures are for the first half of 1942 only.)
While the weights are unfortunately not given in exact form, we can estimate the mean weight at 2.5 kg; the accuracy will be sufficient for the present purpose (it’s probably a slight underestimate, so getting a more precise figure would only make my arguments stronger). Comparing this with the CDC data which I used, we see that it is between the 3rd and 5th percentiles (closer to the 5th percentile).
How should we treat the role of premature births in this weight data? The prematurity rate increased dramatically in the first half of 1942, reaching 41.2%, and then fell to normal levels – 6.5% – in the second half of 1942. While this fall may be partially the result of improved food supply, at least among the pregnant population, the initial rise was also the result of premature births to women who became pregnant before food became scarce. This suggests that prematurity rates for a population under sustained food pressure are likely to return to somewhat normal levels. Therefore, for the purposes of inference to Polish Jewish populations, it would be desirable to separate out the effect of the increase in premature births. (Another reason to do this is the high mortality rate among premature births – 39% in 1942 at the institution discussed in the paper under discussion, and as high or higher at other institutions. As deaths fall out of the population to be studied, these premature births would have no impact on date for average weight.)
Fortunately, we do have data for infants carried to term:
Comparing to the CDC data which I used for birth weights, we see that the birth weights for the first half of 1942 lie slightly above the 10th percentile on the CDC data. The birth weights for the second half of 1942 are higher, roughly midway between the 10th and 25th percentiles for females and above the 25th percentile for males.
Similar results, incidentally, were obtained at other institutions:
The average birth weight of boys carried to term in the first half of 1942 was 529 Gm. less than in the last half of 1941, and of girls it was 542 Gm. less. Though our material is not extensive, there is no doubt about the reliability of these figures, inasmuch as an almost equal decrease in average birth weight of children carried to term was observed in other maternity clinics in Leningrad. In the Snegirev clinic, the average decrease was 500 Gm. In the obstetric-gynecologic clinic of the Second Leningrad Medical Institute the average birth weight in 1942 was 410 Gm. less than in 1940.
Now, the birth weight data at term does not include the influence of the normal rate of premature births. (I am presuming that the CDC data include premature births, although I didn’t see this specified.) Thus, we need to make a correction to take this into account. During the second half of 1942 the percentage of premature births was 6.5%, which the paper says “differs little from the normal rate.” As a first approximation, we might assume that the average weight of a premature birth is half that of a full-term birth (this is certainly too low a figure). Then with 6.5% premature births, the average weight would 0.9675 times the average full-term weight. In that case, it would be 2698 g for the population average, with breakdowns of 2724 g for boys and 2670 g for girls. These lie between the 5th and 10th percentiles of the CDC data (closer to the 10th percentile).
A more accurate method would be to use the data on overall weights, even though it is given in range form, along with the breakdown on the number of premature births. If we accept the estimate that the average birth weight was 2.5 kg for the first half of 1942, then using the average full-term weight together with the numbers of full term and premature births, we can calculate that the average premature birth weighed 2087 g. Combining this with the full term birth weights at a typical prematurity rate of 6.5%, we get an average birth weight of 2743 g. This corresponds to just slightly below the 10th percentile CDC weight data.
If we apply the first method to the data for the second half of 1942, i.e. taking 96.75% of the full-term birth weights, we get 3095 g for boys and 2796 for girls. The figure for the boys lies slightly below the 25th percentile of the CDC data, while that for girls lies somewhat above the 10th percentile. Using the more accurate average premature weight of 2087 g, we get 3127 g for boys and 2838 g for girls.
Thus, in the most severe portion of the famine, during the first half of 1942, once the impact of the prematurity rate is separated out, Leningrad birth weights were approximately on the level of the 10th percentile CDC birth weights. Including the high rate of prematurity, they were between the 4th and 5th percentiles. For the second half of 1942, when pregnancies that began after the onset of the siege were the norm, birth weights were between the 10th and 25th percentiles of the CDC birth weight data. Muehlenkamp’s assumption that CDC 3rd percentile female weights are too high to represent the weights of a starved population is decisively refuted in this case.
Let me take the opportunity to repeat: Muehlenkamp’s assumptions on weights – of children or adults – rest on no weight-data whatsoever.