Statistical Trickery in Pregnancy

I know statistics aren’t people’s favourite thing in the world but having a basic understanding is important in understanding research and how the study authors came to the conclusion that they did. You need to understand this so that when you are trying to question whether something is hospital policy, or whether it is evidence based, you can be confident in your conclusions. You will then be more likely to have confidence in the research, or indeed scepticism of the research, if you better understand the quality and content.

To a certain extent statistics can also be manipulated, squeezed, tugged, and bent over backwards, to get the results that the authors are hoping to conclude. So you need to be savvy. They can also be presented to you in a certain way to emotionally manipulate you and persuade you.

So what do you need to know?

Let’s talk about P-values

You will hear people talk about results being statistically significant or there was a significant difference between x and y. The significance level of the P-value is usually set at 0.05. If you look at the results and notice that the P-value is greater than 0.05, such as P-value = 0.08, then the results are not significantly different. If you notice that the P-value is lower than 0.05, such as P-value = 0.0001, then the results are significant.

Significance means the likelihood that something has not happened by chance. The lower the P-value is, then the greater the likelihood that the difference found has not happened by chance. This means that we can be pretty certain there is a relationship between two variables. A value of 0.05 (the significance level) is commonly taken as a threshold. For example, “there is a significant difference between rainfall levels when dark clouds are in the sky, compared to white clouds (p=0.0001).”

The larger the sample size in the study, the more likely that the conclusion is accurate. If the sample size is small, it may be that the study is underpowered and the conclusions are, therefore, not valid.

What are randomised controlled trials?

Randomised controlled trials are known as the “gold standard” in research. It is where participants are randomly allocated to the experimental group or the control group.

You then need to take a good look at the control group and ask WHAT intervention they are receiving or WHO they are.

For example, let’s cast a critical eye over The ARRIVE Trial (Grobman et al., 2018). The ARRIVE Trial has become famous for concluding women are at lower risk of needing a caesarean section if they have an induction at 39 weeks, as compared to waiting for spontaneous labour.

“Induction of labor at 39 weeks in low-risk nulliparous women did not result in a significantly lower frequency of a composite adverse perinatal outcome, but it did result in a significantly lower frequency of cesarean delivery.” 

This conclusion has been influential in increasing induction of labour rates at 39 weeks in nulliparous women (those that have not had a baby before) (Wood et al., 2023). For anyone that has had a friend, family member, or indeed you may have had, an induction of labour, you will know that it is no mean feat and often results in birth trauma and a dissatisfaction (to put it lightly) with the experience. So the results and impact from this study are a BIG deal.

Let’s look closer at our samples. We know that the induction of labour group are going to be on a labour ward having a highly medicalised birth. But who are they getting compared to? We actually don’t know for sure. It does not specify. We can only assume that they are getting compared to a hospital birth condition. We do not know how many women, in either group, accessed an epidural, known to increase your risk of an assisted delivery (Antonaky & Papoutsis, 2016), how many women ended up having membrane sweeps, how many birthed on a bed, on their back, or how many accessed a birth pool. We know nothing about what antenatal preparation women had done, whether they had a doula, whether they had continuity of carer, indeed whether they even had a midwife or whether the whole process was under an obstetrician.

Basically, the results of this study may only be valid when applied to women who have not had babies before and who are birthing in an obstetric unit. I would question the applicability to women who are planning a home birth, or birth centre birth, known to improve outcomes (Birthplace in England Collaborative Group, 2011), whether they had continuity of carer, known to improve outcomes (Sandall et al., 2016), what pain relief they used, and whether they planned and trained, for a natural birth experience, with good antenatal education and hypnobirthing practice.

I also noticed this in the study:

“A specific induction protocol was not mandated for women who underwent induction in either group.”

So basically, there are potentially big differences in what the actual intervention was in each case?

This is why you cannot take things at face value. You need to dig a little deeper and unpick. If something doesn’t sound like it makes logical sense, that’s because it probably isn’t and attention needs to be paid to the details.

How statistics are presented

This is really important to think about. How statistics are presented has a huge impact on how you might feel about the figures. I would suggest that they often get presented a certain way for a reason.

Let’s look at a common example people hear around 38 or 39 weeks pregnant. “The stillbirth rate doubles after 40 weeks so you should get an induction if you go overdue.” The whole “overdue” thing is another conversation. This is of course alarming. It’s meant to be. Certainly, when I heard it, for the first time ever, in my late pregnancy, I was concerned. I assumed it meant at least a 40% chance my baby would die in my womb unless I had an induction. With this concern, and a bit of common sense, my husband and I sat down to find the research and get some real answers.

Technically yes, the stillbirth rate doubles BUT from WHAT to WHAT. For low-risk women the chance is:

39 weeks = 0.14 per 1000;

40 weeks = 0.33 per 1000;

41 weeks = 0.80 per 1000;

42 weeks = 0.88 per 1000.

(Muglu et al., 2019)

So now we see what small numbers we are talking about. Suddenly, why a high risk intervention, like induction of labour, would be suggested, gets a little more confusing.

I find it helpful to look at these sort of figures as a percentage. To calculate a percentage you divide x by y and multiply by 100.

e.g. 0.14/1000 x 100 = 0.014% chance of stillbirth at 39 weeks.

I also find it helpful to flip things around and look at the possibility of something not happening.

e.g. There is a 99.986% chance of not having a stillbirth at 39 weeks.

Flipped around, the justification for a high risk intervention, such as induction of labour, looks a lot smaller.

When you hear something like “You have a 1 in 400 chance of this terrible thing happening to you and your baby,” it sounds alarming. We all know at least 400 people. You probably have more Facebook friends than that. However, 1 in 400 is a 0.25% chance of the bad thing happening which is a 99.75% of it not happening. Suddenly, not so alarming.

However, for some people the 0.25% chance of a horrendous outcome is going to be too much and they would rather take the intervention, at any cost, to reduce this risk. This is why informed consent is so important to empower you to make your own choice.

You can do this exercise with anything that gets thrown at you. Write the numbers down and look at them in different ways. Then, if you’re feeling really savvy, have a careful look at where the numbers come from.

Remember, never take anything at face value.

This is just a quick snapshot look at statistics and research, so let me know if it was helpful and if you would like more similar content.

Soraya x

Check out my main website

Click here

References

Antonakou, A., & Papoutsis, D. (2016). The effect of epidural analgesia on the delivery outcome of induced labour: a retrospective case series. Obstetrics and gynecology international, 2016.

Birthplace in England Collaborative Group. (2011). Perinatal and maternal outcomes by planned place of birth for healthy women with low risk pregnancies: the Birthplace in England national prospective cohort study. Bmj, 343.

Grobman, W. A., Rice, M. M., Reddy, U. M., Tita, A. T., Silver, R. M., Mallett, G., ... & Macones, G. A. (2018). Labor induction versus expectant management in low-risk nulliparous women. New England Journal of Medicine, 379(6), 513-523.

Muglu, J., Rather, H., Arroyo-Manzano, D., Bhattacharya, S., Balchin, I., Khalil, A., ... & Thangaratinam, S. (2019). Risks of stillbirth and neonatal death with advancing gestation at term: A systematic review and meta-analysis of cohort studies of 15 million pregnancies. PLoS medicine, 16(7), e1002838.

Sandall, J., Soltani, H., Gates, S., Shennan, A., & Devane, D. (2016). Midwife‐led continuity models versus other models of care for childbearing women. Cochrane database of systematic reviews, (4).

Wood, R., Freret, T., Clapp, M., & Little, S. (2023). Rates of induction of labor at 39 weeks and cesarean delivery following publication of the ARRIVE Trial. JAMA Network Open, 6(8), e2328274-e2328274.