Journal Club 1- Breastfeeding and childhood cancers

tallaght01

So, this may or may not work, but we'll see.

I was reading a paper this month (The only paper I've looked at this month ) while I was doing some random reading, so thought I'd use it to get the journal club ball rolling.

I've tried to keep the subject matter such that it could potentially interest a wide variety of people.

The good news, too, is that this paper has free online access, so there's no copyright issues.

The paper is:

Ortega-Garcia JA. Full Breastfeeding and paediatric cancer. Journal of
Paediatrics and Child Health 2007;44:10–13

and it can be found at:

http://www.blackwell-synergy.com/doi/abs/10.1111/j.1440-1754.2007.01252.x

Just click under the abstract for a PDF or HTML full text.

The study was caried out by a Spanish group, who wanted to examine whether breastfeeding has a protective effect against the development of childhood cancers.

They conclude that it does, as long as there is at least 8 weeks continuos exclusive breastfeeding. The effect is related to the time spent breastfeeding.

They have printed their odds ratios in table 2.

To come to this conclusion, they looked at 187 patients with childhood cancer, and the same number of controls.

They questioned their parents about how long each child was breastfed for, and the exclusivity of breastfeeding.

They report that the children with cancer were less likely to have been breastfed, or were breastfed for shorter periods.

So, to make this clinically relevant, let's look at the practical application....

You're a healthcare worker/scientist/student, so you "know about these things". A patient/friend/relative has been reading in the metro that breastfeeding protects against your kids getting cancer. They want to know if it's true. This person is about to give little Timmy a bottle of SMA because "he's starting to bite, and sore nipples are no fun, I can assure you".

So, what would you tell them?

Does this study add anything new to the information you'd give them?

To me it doesn't add a huge amount, and I haven't mentioned it to parents as of yet.

In these type of trials, recall problems are a huge reality. You can see in table 1 that some of these kids haven't been breastfed in over 10 years, yet we're asking mums to recall exactly how long they breastfed. This recall ability is relied upon to the extent that they claim a 0.97 change in your odds ratio for every week of breastfeeding.

The "people who know about these things" tell us that, even when we think we remember exact details from many years ago, we're not usually as accurate as we think.

IN fairness, they have taken steps to reduce recall bias...but still.

I would love to see a prospective study.

The odds ratios themselves aren't earth shattering, and I'm not sure how much of a difference breastfeeding is likely to make to childhood cancer rates on a population level. Maybe someone brainier than me could give us an idea.

My stats are poor as hell, but the statistical methods look OK to me in this paper, but I'm open to correction.

So, what do others think? Should I be printing up more blackmail posters for the nursery at work? Or should I hold fire?

Reading papers is my big weakness, so I'd be very interested in the views of others.

charlieroot

Nicely written paper overall. Concise which is always good So my quick thoughts:

Results:
There seems to be quite large differences in the mean and median for breast feeding duration in both cases and controls. Could this affect the results? Is it indicative of high variability in the data?

From wikipedia; there is roughly 650,000 births each year in Spain using the birth rate from 2006. There are about 900 cases of PC p/a in Spain meaning an incidence of about 13 cases per 10,000. Can we work out the change in incidence if all mothers breast feed say for example at least 16 weeks? I think it would give a good indication of whether this is clinically significant or not.
Anyone else think this is a calculation the researchers should have done?

Discussion:

Possible sources of bias:

Was there a higher number of first born cases or controls? Any significant differences? (Just a random thought of mind... could be / probably is totally irrelevant.)

Was the child diagnosed by the time of the second birth? Were the parents somehow "extra careful" with the next child?

On recall bias - could misplaced guilt of the parents perhaps convince them that they didn't take care of the child who developed cancer as well as the child who didn't and perhaps affect their recall?

Single interviewer?

Since the siblings were being used as controls - could genetic factors not confound the results?

Random Thought:

Could the positive effects of breast feeding actually be an inverse negative effect of something in the formula for bottle milk?

Chunky Monkey

The PEH included information on birth weight, gestational age, delivery type, socio-economic status (net family income), mother's educational level, mother's smoking habits during pregnancy, and mother's current working status (employed or employed).

How come no mention on whether someone else in the family (eg aunt/uncle, grandparents) had the cancer or does using siblings as a control void that?

You're a healthcare worker/scientist/student, so you "know about these things". A patient/friend/relative has been reading in the metro that breastfeeding protects against your kids getting cancer. They want to know if it's true. This person is about to give little Timmy a bottle of SMA because "he's starting to bite, and sore nipples are no fun, I can assure you".

Why isn't she pumping!

littlebug

Very interesting.

I'd like to see some qualitative stuff on the reasons why the (case) mothers didn't breastfeed or stopped breastfeeding earlier.