Am I reading the CDC Data Wrong?

General Questions
#1

Data from http://205.207.175.93/Vitalstats/TableViewer/tableView.aspx

Supposedly there’s a correlation between tobacco use and low birth weight… however, I DONT SEE IT in the data. Am I reading this wrong or what?

(Note I’m NOT trying to backup tobacco use while pregnant, or even at ALL. I just need help reading the charts correectly)
(Approx. Rounded)

Gestation Period - http://oi42.tinypic.com/wgqfk3.jpg

[Total] 1 Smoker to every 8 non smokers
[Under 20] 1 Smoker to every 10 non smokers
[20-27] 1 Smoker to every 6 non smokers
[28-31] 1 Smoker to every 8 non smokers
[32-35] 1 Smoker to every 7 non smokers
[36] 1 Smoker to every 11 non smokers
[37-39] 1 Smoker to every 12 non smokers
[40] 1 Smoker to every 12 non smokers
[41] 1 Smoker to every 11 non smokers
[42+] 1 Smoker to every 7 non smokers

Birthweight - http://i39.tinypic.com/mwr3pl.png

(Weight, Smoker/NonSmoker)

499 1/6
500 1/10
1000 1/7
1500 1/7
2000 1/6
2500 1/6
3000 1/9
3500 1/14
4000 1/17
4500 1/22
5000 1/20

Thanks!

#2

I don’t know how exactly you are reading it, but it clearly shows SOME sort of correlation to me. How statistically significant is not something I am going to put the effort into determining though.

For both data sets the overall ratio is closer to 1 smoker to 9 non-smokers than 8 non-smokers. For the sub-groups, if the ratio is 1 smoker to 6, 7 or 8 non-smokers, then that means that smokers are over-represented in that group compared to the overall ratio. Similarly, if the ratio is 1 smoker to 10, 11, or 12 non-smokers, that means smokers are under-represented in that group compared to the overall ratio.

Looking at the data, the smokers are over-represented in the lower birth weight and shorter gestation period groups. There are a couple of exceptions at the extremes, but those are also smaller samples.

#3

A simple way to look at it.

Smokers represent 0.059 of total births. (236K/4000K) Low birthweight is defined as under 2500 grams. Smokers represent 0.086 of all low birth weight births. (28K/325K)

Parsing the data one more way. The risk of low birth weight given being in the smoker column is about 12% compared to the risk of low birth weight being in the non-smoker column which is significantly less, about 7.4%.

There are of course more sophisticated ways of handling the data but that’s enough to give a basic picture.

#4

Here are a couple articles that explain the effects of smoking on fetuses and newborns. The first is easier to understand.
CDC, NIM via PubMed
The low birth weight is often due to preterm delivery.

#5

Moved MPSIMS -->GQ.

#6

And also often due to being small for the gestational age (SGA) as the placenta is unabe to provide enough nutrition, causing intrauterine growth retardation (IUGR).

The op, if I understand correctly, was not asking whether or not smoking causes these problems, but how to read the datasets. The IUGR/SGA bit is easy to see by just looking at the 40 week only columns. The mothers who were smokers who delivered at 40 weeks gave birth to low birth weight (under 2500 grams) babies about 3% of the time, compared to only about 1% for the mothers who were not smokers.
Caiterz, looking at it the way you organized it, what seems unclear? Births 36 weeks and over (“near-term” and term) and not post-dates (also an abnormal state) ran about 1 out 11 as smokers (rougly 9%) and those 35 weeks and under (preteerm births) ran averaging about 1 out of 7 (roughly 14 to 15%) as smokers. That’s a pretty big difference (of course a real statistical analysis needs error bars and confidence intervals but given the large n I think we can dispense). By birth weight you have smokers as roughly 1 out of 6 (16 to 17% ish) births under 2500 and anywhere from 1 out of 14 to 1 out of 22 births 3500 grams and over (5 to 7% ish range).

What are you not seeing?