You're quoting from a detail that had nothing to do with the survey although as it happens it's pretty damning of the conclusion you're advancing - that there are no dangers worth a warning.

"IVF babies twice as likely to have heart defects and four and a half times as susceptible to digestive tract disorders... 3.7 times the risk of bowel disorders and four times more likely to have a cleft lip... a survey of births in Australia showed a two fold increase in the risk of major birth defects..." and so on..."


Given the high failure rate 20,000 was probably all births from about 100,000 attempts. Either way they're clearly convinced that people need to be warned about the dangers.


Obviously the most accurate research would access official statistics - their role was to apply a rigorous scientific analysis thereof and reach conclusions thereon. Let's be honest - persons considering such a procedure would be more interested in what impartial experts have to say than in the guesswork you lot are all straining to advance a Mac forum.

km

You're right, what I posted was complete rubbish about them using all IVF numbers and so they were weighting the study including all of the startup IVF . The study used 1997 to 2003. My mistake. But since you posted a link to the original paper I read it.

All I can say is that the conclusions in the paper might be true but what you think they concluded is false.

They just got over the threshold so they think that IVF should carry a warning. Sure, since everything else carries a warning and anyone submitting to IVF has to sign so many an extra one will do exactly nothing. As it should since this effect they are measuring is so small.

Between their controls and their samples they have to use multiple logistic regression to normalize a few things. They have to take into account maternal age, ethnicity, education of parents, income and quite a few other things. Why do you think they have to negate those effects? Answer: because all of those factors contributes more to birth defects than the thing they think they are measuring, the actual use of IVF. Since those other factors carry more effect should we warn people who are not white, rich, and educated against giving birth at all? You might have some trouble with that one.

There are two other glaring problems which they mention in the conclusion but sort of sigh about, they cannot negate. One is ascertainment bias problem. I know for a fact, as you should, that an IVF baby is going to be looked at post natally much more than any other baby. Because the baby (using their data) is born to white non hispanic, upper income, educated parents. They will bring their kid in for minor hyper spadious while a poorer child it is ignored. Or a slight cleft pallet. They admit they cannot control for this although this ascertainment bias could account for all of their elevated numbers.

Two: They do not discern why this effect could be happening. I could come up with several which would not make me a know it al nor would any of them make me not want to try IVF.

nb: Your conclusion that it is multiple births that is the casue is proven false in the paper. That is about the only thing they could conclude definitively.

One last note directly from the paper:

Good gravy that is an amazing conclusion. But it is the lot in life of epidemiological studies like this using humans as data. Let's parse that three part conclusion by the authors:

1) Underlying infertility: Well gee, maybe. Given that the parents are presently infertile do you think they would risk this 1.5% extra chance of a birth defect?

2) Small numbers: Holy cow are they small. The N is pretty small but the numbers of actual children with IVF and birth defects are really really small and they do not give actual p-values. Which leads to:

3) The conclusion could be based on chance. So small that one baby detected with a very minor birth defect but because the mother is loaded they report and fix the cleft palate which is just barely cleft while just one baby missed who is poor and black and cleft palate screws the whole pooch on this study. This is one of the major flaws of the study. They don't figure out the proper p-values or at least I just cannot understand what they are doing when they produce p-values that are < 0.00 for traits of the participants but don't create p-values for the results themselves. My only conclusion is that the p-values of the conclusions support random chance as much as their conclusions.

If this was a study using mice the whole thing would be thrown out as nonsense because the error bars and standard deviation is much larger than any conclusion they think they made.

So you were right about my last post but I agree with your sentiments just the wrong target. This study is complete rubbish. But it was an interesting read. I love to tease epidemiologists that they are not real scientists and that Political Science has more science in it than what they do. I'll use this study. Thank you.

Your presumption here is that the one event is unacceptable ... I don't think that holds true for our society today.



Noted. I disagree. Although I wish you had presented your position thusly in your opening post as it might have generated a more focused and interesting discussion rather than one mainly about statistics ...

Not really - from the paper:

"In this study, 23.4% of all twins occurred following ART. Thus, ART might contribute to the risk of major birth defects both directly by increasing the risk of defects among singletons and indirectly by increasing the occurrence of twinning which is a strong risk factor for many types of major birth defects (Li et al., 2003; Tang et al., 2006). Even if the additional impact of any risk posed by ART is negligible among the already high-risk multiple births, the strong association between ART, as practiced in the USA, and multiple birth should nonetheless be considered as another pathway through which ART might indirectly contribute to birth defect risk."


That was the whole point of applying multiple logistic regression adjustments - these guys are not amateurs but leading experts in the field which is why their conclusions are being taken seriously where it counts - such as with the Human Fertilisation and Embryology Authority (HFEA). Accordingly HFEA has emphasised the dangers of multiple births from IVF on its web site as a matter of statistical fact for example that cerebral palsy is five times higher for twins and 18 times higher for triplets than among other children.

km

I guess I agree with your first point which I did not understand. That an increase in IVF will increase twins and that this may be the cause of all of the additional birth defects. I got that. I thought you were arguing that IVF caused birth defects. It doesn't. The ever so slight increase of 1.5% could be caused by twinning.

You then say these guys (although the first author I know is a women) are not amateurs. Correct. That is why you should read what they say in their conclusion which is that all of what they are reporting could be because of low numbers, or by chance. Or it all could be associated with multiple births. But again, the numbers are really too low to conclude much of anything except when couched in terms of "might" and "could".

They are not amateurs but they spent a lot of money on this study and need to publish thus we get papers that are very inconclusive unless read by people who do not understand the statistics and conclude that something worth talking about and their worst fears are being realized.

Your argument from the beginning would have been clearer if you presented this early on. Fear of man messing in God's domain or "unnatural" acts is a normal response to something new and thus "abnormal".

I disagree that IVF is in any way unnatural. There is no use arguing that point as it is to discuss how many angels on the head of a pin. And this paper in no way raises the issue of "dangerous" when speaking of 1.5% increases of minor birth defects during child birth. By your criteria breathing is dangerous and thus we should be suing anyone who forces a dutch oven upon us. I think I need to sue my wife right now.

Considering that she holds all of your assets, it could be possible. I know when I got married, everything became hers and ours - and ours is also hers by default.

Not particularly... playing God with a biology set is what's dangerous and as a result children are being born with horrific disabilities.


phOhyeah... some crank comes along with a test tube after 350 million years of births by natural selection of the strongest sperm and you buy his random sperm experiment as a natural process? In vitro means taking place in a test tube outside of a living organism - what's so natural about that?

km

It does. HFEA say that multiple births are the biggest risk, not the only risk. Significant risks to both parent and child are caused by the procedures themselves.


I believe the increase is 1 per cent over the normal rate of 2.5 per cent which is a 40 per cent increase in risk.

km

On "natural," been there and done that with KM. Don't try it . . . you'll get nausea from the spin.