Breastfeeding & health care experiences: study design -next steps

In our chat this week, the Breastfeeding and Healthcare Experiences group decided their topic and also had a think about specifics: What exactly would we do? What numbers would we aim for, what techniques would we use, and so on?

This blog post focuses on our main topic, and these notes are intended to help the working group tasked at turning our chats into a draft protocol.

The group chose the topic:  What incorrect or unhelpful pieces of advice do HCPs give to breastfeeding mothers?

(They also chose a side project, more on this soon…)

Should we focus solely on the negative aspect?

The question above focuses on negative advice.

However the general feeling in discussions seems to be that it would be best to include all experiences for the reasons outlined below. Does the working group agree with this? What should we be asking instead?

Two questions:

  • shall we look at both positive & negative comments?
  • if so, how can this new question be presented as a research question that can be answered?

Comments from our recent chat:

PSG F: are we going to refine our research question? A number of people have pointed out that we should investigate both negative and positive advice. If we do, there is a potential mismatch between our question and our data (open forums).

PSG AH: I am worried that [this question] seems to bias that we are automatically assuming HCPs are giving out bad or unhelpful advice, but some variation on this would be good

PSG AE: I suppose with the terrible rates of being and the knowledge that we’ve all been given bad advice at some point. We are interested in studying what things are said by HCP that are exacerbating this and what do they need to stop saying?

PSG G: I think we can easily also ask people – who helped your BF journey? For me personally, I haven’t had much specific good/bad advice from GPs etc but there was someone in the NICU who I credit with me breastfeeding at all. And some positive advice from my HV. So I’m keen to highlight the good as well as the unhelpful

From personal experience, people post more frequently about the unhelpful advice than helpful (eg, do I really need to pump and dump while I take this medication?). Of course this would happen with a survey as well but on the flip side, posts about positive advice might also be less specific (eg endorsements about a HV or GP). Also I see more endorsements of private HCPs (I need a tongue tie cut ASAP, is their a private MW or lactation consultant anyone can recommend?)

So if we want to explore positive advice (which I think we should) we need to ask people more directly rather than analysing historic forum posts.

PSG H: I would lean toward rephrasing [this question] to ‘what advice’ rather than going straight to the negative.

I’m now curious to know the opposite – what correct & helpful advice did mums receive?! For me, i can never thank the midwife enough who helped me through suggesting pumping pink milk, syringe feeding, nipple shields, & gave me tips of how to ward off the beginnings of mastitis. If it wasn’t for her I don’t know how long I’d have lasted…

I think others have suggested that advice can also be interpreted in different ways (positive by one mum, and negative by another)

PSG E: How do you define a negative word? 😉

PSG T: maybe [this question] should either look at the good advice and not the poor advice (flip it around so it is positive) or could it focus on the contradictory advice given? An example of which would be that I was told by a Dr that my iritis was solely due to being run down because I was breastfeeding, my Optician reviewed me and said that it possibly had something to do with it but could look at the bigger picture and said I was run down because of a flare of my seronegative arthritis, stress (due to doing a million things at once), sleepless nights, working and raising 2 children and the effect that could have on a mum?

PSG I: I’m still very uneasy about specifically looking at unhelpful advice. I think potential for negative press all round from HCPs themselves and for PSG could be quite high. Not least because in picking unhelpful advice you’re making the assumption (I think) that this affects Breastfeeding journeys more than helpful advice. Do we know this? Has anyone looked in the literature to see whether anyone has previously correlated HCP advice type and Breastfeeding outcome?

Has anyone looked to see if type of HCP advice (good/bad) and BF success/outcome has been studied? It might be that good advice leads to better outcomes more often than bad advice leads to bad outcomes. Make sense?! In which case we’d be wanting to really focus our efforts on exactly what that GOOD advice is rather than looking at the poor advice.

PSG V: I think HCPs need to know when something they advise has the potential to be dangerous or has the capacity cause tangible clinically negative results.

2 things I was told were wrong. And definitely so. Clinicians are used to being told what may cause harm and how to change. I feel quite strongly on this.

PSG W: That’s a very interesting perspective. Offering boiled water and infrequent poos ubder 6 weeks are examples of this.

PSG V: Being told not to breastfeed with mastitis and being told to make my tongue tied son ‘understand what it was to be hungry’ so not to feed on demand and keep to 3 hours. He was very hungry already :/

In both cases I luckily asked people with more specific training for advice.

PSG X: Thx is what I’m interested in too

Who would we get the data from and how?

Broadly, there are three main suggestions (with variations on them!)

  • use our own experiences
  • use posts already in existence (Facebook groups / online chat forums)
  • invite submissions (from breastfeeding networks or …? )

Using our own experiences

PSG AH: But wonder if we can pool our own responses into the system too – it would be a shame to focus completely on others and miss the expertise and experience that group members also bring

I have concerns about the ethics of investigating closed groups but we could use our own reflective experiences.

PSG N: Petra said:

“I’m biased as I love reflexive projects (where you bring your own experiences in) but I think that could be very powerful indeed. And when it comes to convincing other HCPs having that real world, personal experience really engages them.”

NOTE: There is more discussion on this in our last chat with Petra Boynton

Using data from groups – including discussion on ethics

This very project was originally inspired by the number of posts in BFing support groups with mothers reporting advice given to them by HCPs that ranged from very positive to non-evidence based advice likely to damage to the BFing relationship and even the downright bizarre.

Yet – we are all part of BFing support groups ourselves – we are the target group! – so we understand very well that although there is rich data here, there are ethical issues with using it. We don’t want to do anything people in those online communities would feel uncomfortable with.


Question for working groups:

Do we want to work with the data from groups and/or online forums? If so, what’s an ethical way to do this? Is it feasible time wise?

PSG A: Definitely need to think about ethics of data
Mining from Facebook groups, etc
Ie is it ethical?

PSG C: Looking at comment boards, we could assign people to a board and a time period, going progressively back in time as per number of volunteers.

PSG K: I’d list the groups and assign people to look for keywords within facebook groups, logging each original post with date, location and name (though not sure how this would work ethically) so that poeople who post in more than one group dont get picked up twice.

The issue type discussed would be logged, age of child, outcome etc,

We’d also go through the comments because people often reply with their own experience and the same thing be logged. All with permanent links to find again…thats my start

Examining posts in large breastfeeding groups for mention of interactions with HCPs and then scoring them into categories (ie type of bad advice), age of child, and type of HCP.

PSG D: Thematic analysis will probably be useful with broad ideas like sleep and medication and refining down to more specific directives.

PSG G: We could either go searching for HCP related posts in the groups we’re in and then contact those people to either get more info or ask if we can use their experience. Or we could put a shout out in those groups to ask people to contribute their experiences specifically for the research. We can also easily put our own experiences in.

PSG S: I think forum posts will throw up negative and positive experience from interlocutors. The mother asks do I really need to pump and dump and people reply either their mw/hv/doctor said no (positive) or they said pumping and dumping was necessary (negative) but the mother found her answer elsewhere. But we may be able to do more in depth analysis of experiences of mothers who have received positive help and therefore managed to continue.

PSG K: maybe we can have a post in each group saying what we’d do, and get in touch with each person to ask if theyre okay with it being logged . Its massive work but it’s personal and a lot more likely to get accepted?

PSG M: Perhaps it could work in 2 parts…firstly we scan the groups for what we know to be incorrect advice, then based on this we make up a poll to run in each group with each piece of advice as a poll option and ask people to participate in which they have been told. Gets round the ethics issue possibly?

PSG A: Are we focusing on FB? or other online forums too?

PSG K: Would we use FB solely or sites like mumsnet etc, theyre massive
PSG N: What’s the difference between FB group posts and MN posts in practical terms?
PSG N: e.g. MN are anonymous, FB not
PSG C: Mums net is open, some Facebook groups are closed?
PSG D: MN is openly available and searchable, FB groups are not.
PSG O: I don’t use MN. Do you have to join individual chats to post? Or do you ‘put up’ questions?
PSG D: MN have previously given consent for posts to be incorporated into research.
PSG D: MN has a corpus of data going back nearly 15 years.
PSG P: Does MN tend to attract a certain ‘type’ of mum? I.e. are the majority of users fairly typical of the population, or is there a specific dynamic over-represented? Then again, would that matter anyway so long as there’s a large number who can answer our question?! & actually is a group with a lower percentage of breastfeeding mums more helpful on the assumption that they were given worse HCP advice?!

PSG P: Then again (following on from my last comment) a group with less breastfeeders may mean people who don’t realise the advice they got was unhelpful/created more problems, & so they may give less accurate answers (or feel defensive and not reply)

PSG N:  you have to join to post, but you don’t have to join to see everyone else’s posts.

How many posts?

Sophia Collins: This may be an impossible question to answer (and we aren’t going to be tied to the answers), but ballpark figure, how many posts about interactions with HCPs would we want to audit? 100s? 1000S?

PSG B: I was thinking this, this reminded me of at the APPG when Amy Brown said she started a survey and it got shared so much she ended up with 18000 responses to look through ?

PSG F: I don’t think we have time to audit 1000s, unless someone has a specialised software package and experience doing this type of analysis.

PSG B: me neither and statistics in research make me personally want to end it all, standard deviation in particular – although someone might really want to do that stuff!!

PSG O: Getting general info for this group skimming threads, I manage a hundred or so in 15-20mins

PSG N: for the others who weren’t part of that quick review, would you be able to describe the process of what you did?

PSG O: In depth and logging maybe a 100/hour? I skim read looking for GP/MW/HV mentions on bfing groups. Had written a list of the topics likely to arise and then scored them down as they came up.

Ethics of mining data

PSG D: I am very uneasy about mining existing groups from an ethical standpoint but people are often happy to share their experiences when asked, which allows us to reach out to mummy groups which aren’t breastfeeding specific, and that will help us avoid inherent bias.

PSG A: I agree
PSH L: Also agree
PSG K: thats a good point
PSG B: I agree too, I think there needs to be complicity

PSG C: This might be helpful:

Ethics, Privacy and Internet Research

Systematic observation of/ engagement with the research site prior to and alongside data collection and analysis;

Memo writing or other self-reflexive activities that will help you to engage with the process of data construction and to document your observations, personal responses and data selections in a critical way;

Adopting a participant stance: by situating yourself within the research site, you can better engage with ethical issues from the perspective of those who are being researched.

Making early and regular contact with gatekeepers such as site moderators or employed staff. These are the people who are likely to have the best understanding of your research site, how people engage with it and what potential there may be for your research to cause harm to its users. Talking to individual users is important too, where possible: it is very likely that people will have different perceptions of their engagement with internet sites and marginal views should not be ignored.

Linguistic analysis. My understanding of the norms of information sharing, privacy and anonymity in Mumsnet Talk was greatly strengthened by my close linguistic analysis of participants’ interactions. For example, I found that some Mumsnet users achieve a degree of privacy when they construct and address specific in-groups of users, through their use of shared in-group knowledge and linguistic resources.

As mentioned previously I don’t think we should be data mining, I think we should be gaining consent but I think we have to consider that experiences could be triggering in some way and that we take that into consideration somehow.

PSG Z: Data mining without the express consent of those involved is not something I would be keen on doing

PAG AA: Do you mean on FB groups or also on forums like MN (where it’s publically available)

PSG AB: 1-groups who are happy to do so can update their pinned posts saying that any new new comments may be used for research purposes, completely anonymised and if want to opt out can say so on that
2 -old threads maybe retrospective approval but not sure how
3- place a post in groups with admin approval with the question of experiences, people can reply to that post

PSC AC: I posted this link in the previous comments:
Ethics, Privacy and Internet Research
PSG AB: really interesting

PSC AC: I completely understand why people feel uncomforable about ‘mining’, but on open (public) groups, it is a ‘normal’ research practice that hundreds of university researchers use every day.

PSC AC: I haven’t read these in detail, but it might be useful to:
The British Psychological Society guide to internet mediated research

PSG AD: If we were looking at existing posts in groups we could message the poster to ask them permission. We could set up new groups and invite people to come and share their stories or share posts and hope people see and comment. I think if data is anonymised and the group is public then it wouldn’t be unethical. Though a lot of BF groups are private so people feel OK about sharing fully.

PSG AB: setting up new groups is a good diea

PSG AD: Just thinking of the PSG ones! We mustn’t forget we can use our own experiences too, I’m not sure how many people are in the main HCPs science gang group but that’s a start!

PSG AB: how many PSG groups ARE there? it’s like they’re in every topic

PSG AI: There are 8 PSG groups.

PSG AE: there are 39 of us in here

PSG AD: That is a good point, I’m sure there’s some overlap but we could also go to the other topic PSG groups – people might be more likely to contribute if they’re already bought into the concept of parenting science!

PSG AD: that’s just tonight’s chat – I just checked and there are 348 members in the main ‘Breastfeeding and Healthcare Experiences PSG’ group

PSG AE: However are we mostly extended bf’ers? how do we reach the people who had bad advice and stopped feeding early

PSG AE: Sorry just clicked top of screen and didn’t think about this being a subgroup

PSG AD: That is true, we might be more biased towards long term breastfeeding but it gives us a big chunk of our population, then we can go looking for other groups

PSG AF: Yes PSG AE: , where I grew up most of my friends gave up by day 5…& are now extremely sensitive/defensive. But I’d love to study their experiences & what could/should have been done instead

PSG AB: without somehow activating the defensive response -I’d love to know how to do that

PSG Z: I so far have never managed to avoid that reaction sadly. Despite best intentions/efforts

PSG AF: Yes I’m flummoxed as to how to share incredible breastmilk facts & feeding tips (that may help motivate mums to commit with the needed determination & also be equipped with the practical knowledge) without risking being seen as pushy/pressurising/judging…if anyone knows, please tell me! Or should this be our next research question…?!

PSG AG: I don’t personally have a problem with us analysing open forums (like mumsnet), especially if mumsnet has already given approval for their posts to be analysed. If we’re going to aggregate the posts and report back high-level messages, the details of individual stories don’t need to be re-posted. If we do want to include excerpts from people’s specific experiences, we could just get their consent for those. Otherwise, we’re contacting 100s of people for their consent to analyse their public available information.

PSG J: I think the main points for me are

  1. The data is from human subjects and we would be doing ‘research’
  2. Accessing it on closed groups (say on facebook) without knowledge of members could be deemed deception, especially if it’s used for any publication. So, it would be a matter of reconciling those two areas in whatever we chose to do.

PSC AC: All the discussion so far has been about public posts, on open groups that are not password protected.

PSG J: All the breastfeeding Facebook groups are not public.

PSG AG: I can’t think of a single public FB BF group. We wouldn’t be able to use BAPS, BOBAB, BYBAB, BFS&I….

Inviting submissions

Recruitment survey

One suggestion was to recruit via a short survey.

PSG F: Contacting individuals for consent to use their HCP experiences could be quite time consuming. Opt-in consent with a poll or survey would help this.

PSH L: Asking for contributions via survey/questionnaire on FB groups and websites
And ask if the person would consent to a more in depth interview? Mixed methods?


PSG C: This  sounds like questionnaire
PSG AJ: Survey shared on as many people breastfeeding social media pages as possible.

Self reporting survey diffused to breastfeeding groups with the following questions

  • Has a healthcare professional ever recommended using formula while you were exclusively breastfeeding? Yes/no
  • What kind of healthcare professional?
  • In what context: routine appointment including developmental checks and weigh in, experiencing feeding difficulties, maternal health problem requiring treatment, concerns about sleep, other
  • Did they recommend a specific brand? Yes/no
  • If yes which (tick box but I’m not giving them the guts here)
  • How did this make you feel about your breastfeeding experience?
  • How did this make you feel about the HCP?

PSG D: We may be leaning towards what advice target than focusing on incorrect which would make a survey easier because you put everything in there. You could band by age as the survey goes through because (in)correct at 2 months isn’t necessarily so at 20 months.

PSG J: Can I ask why we wouldn’t create a survey or questionnaire to collect information, so having opt in consent…and opt in to being contacted for an in depth interview (if this is also what we will do)? It seems an option to me, and one less is more transparent than looking at people’s comments and experiences without their express consent.

The link would then be promoted and shared by us and by collaborators and admins of the group.

PSG AK: Petra said that both have their limitations, that surveys require more work than you think up front. She also said if it’s the best way to contact your target group to consider it.

PSG J: Thanks, that makes sense. But, we are the target group in many ways and have inside knowledge of what questions need answered I think.

Thoughts on Interviews

PSG D: If we’re doing in depth analysis from interviews it’s a lot more time consuming than categorizing themes from things we’ve read.

PSG A: Definitely – But also offers more rich info/data


PSG B: I was thinking if we focused on language negative words for example, that could just be a tally which would lend itself easily to quantitative research but like you say not exclsively

I think it would be more phrases but we do hear the same anecdotes over and over “no benefit” “you need to stop because” etc

Corpus Linguistics

PSG D: It could be done with a corpus linguistics tool, however I no longer have a license for mine…. There may be reasonable free ones on the web. That way we can pick up keywords and easily look at the phrase they’re contained in.

PSG D: This tool can be used on webpages and returns key word in context

Turbo Lingo tool

If this is the route we want to go down can we get a n expert in corpus lingusitics to come and chat to us about using this method to kick up data?

PSG D: Oooh hello Dr Gavin Brooks of Nottingham! “The finer details of the bigger picture: corpus linguistics in healthcare”

PSG K: Woah never heard of this approach!

PSG F: Gavin Brooks is very well placed to advise. He’s worked with Tony McEnery analysing 1000s of pieces of online feedback left on NHS services. Find that people mostly complained about rude receptionists, not the care they were receiving.

PSG D: It’s really a interesting method of quantifying incidences of words from qualititative data and looking at how words are used. It’s probably moved on loads from when I last did any properly ( ?2010?).

Other comments to consider

PSG AK: If we’re asking people directly we should also ask them if the advice helped or hindered their breastfeeding journey. This is the ‘so what’ question; mums have been receiving good and bad advice, but did make a difference? Did they ignore the poor (or good!) advice and soldier on add they were?

PSG K: I like the idea of asking people to comment but it never has the response because its a forced question, if that makes sense.

PSG O: [Whatever we do] would needed to be worded so it wasn’t biased and yet also not ambiguous!


Leave a Reply