Bedsharing was assessed at 2 months on a scaled variable with answers including "never" (55.8% of the sample), "once or less per month" (10.7%), "one to four times per month" (14.2%), "two to three times per week" (7.41%), and "four or more times per week" (12.2%). As you can see, regular bedsharing wasn't incredibly common, but some form of at least occasional bedshared was found in nearly half the sample. For the analyses, the researchers included both a dichotomous variable on bedsharing (never vs. any) and a trichotomous variable focused on frequency (never bedshared; some bedsharing; frequent bedsharing). Notably, the 'some bedsharing' infants included everyone who bedshared except the group who bedshared four or more times per week, who comprised the 'frequent bedsharing' group.

Notably, researchers also assessed "nighttime parenting" more generally at 2 months. The measure comprised of a questionnaire based on 10 different types of comforting provided to infants and asked parents to rate the frequency in which they used any of these over the previous two week period. The items were found to represent two different types of comforting when the baby cries in distress: proximate (including items such as holding the baby or rocking the baby) and distal (including items such as showing the baby something to look at or changing the baby's position).

Finally, attachment security was assessed at 14 months using the Strange Situation Paradigm, a well-validated, objective measure of attachment security. This provided two variables: attachment status (secure, insecure avoidant, and insecure resistant) and strategy (organized or disorganized). This is based on recent research that has found the disorganized attachment is better reflected as a strategy as some children can be securely attached but with a disorganized strategy. The authors also quantified attachment security using a validated continuous measure based on the SSP which provides continuous scores for both secure attachment and disorganized attachment.

In addition to these primary measures, the authors controlled for maternal age, maternal education, parity, maternal depression, breastfeeding at 2 months, breastfeeding duration in months, perceived infant temperament, and crowding in home. Thus, the reported analyses controlled for these variables when assessing the relationship between bedsharing and attachment security.

Based on numerous analyses, the authors found that solitary sleeping (i.e., never bedsharing) was associated with an increased risk for a lower secure attachment score using the continuous measure and at increased risk of having insecure attachment, with a specific higher risk for the insecure-resistant subtype. There were no relationships with disorganized attachment. Of note, however, when the frequency of bedsharing was examined using the trichotomous variable, there was no dose-response found, with the primary differences occurring between the 'solitary sleeping' group and the 'some bedsharing' group.

Now to the inevitable questions…

So… what do these results mean?

The take-home from the research is if it is taken at face-value, all the people arguing that bedsharing causes later social problems need to take a step back as not only is there no evidence supporting their claim, but now we have evidence flat-out contradicting it. The type of insecure attachment that solitary sleepers are at higher risk for here – insecure-resistant – is associated with more "clingy" behaviours later in childhood. Thus, the act of trying to force independence at nighttime on children to make them more independent may actually backfire. As the researchers state,

"Our results therefore provide some support for the theory that children who do not have nighttime access to their parents are in fact more likely to be "clingy" (resistant) later on than are bed-sharing children, contrary to some popular conceptions. (p8)

However, as much as I think many people would like to say that this is "proof" that bedsharing increases attachment security, the research itself is definitely not strong enough to make that claim. The research is preliminary, thus so is their evidence and it needs to be replicated and remaining questions (such as the reason for the lack of the dose-response) need to be further examined. That said, it certainly does warrant dissemination and discussion given that so many of today's "experts" are out telling families that they should not bedshare under any circumstance.

What happened to the measure of nighttime care? How does that fit in?

Proximal nighttime care was found to be associated with bedsharing. Specifically, frequent bedsharers engaged in the most proximal nighttime care activities followed by those who bedshared sometimes and then the solitary sleepers. There were no differences in the groups in terms of distal care and it was not regularly engaged in by anyone. Now, it would be logical to think that it was this was the crux of the relationship, but oddly proximal nighttime care did not have the same predictive power as bedsharing. This suggests there is something else happening on top of the responsiveness that is being assessed using proximal nighttime care or that this particular measure of nighttime care is failing to ascertain the entire or true nature of responsiveness at nighttime and bedsharing may better capture this responsiveness. (I'll get to this more below.)

Why might there not be a dose-response? What might be another explanation for the results outside of solitary sleep increasing risk?

First, there may not be a dose-response because the primary relationship obtained herein isn't "real". That is, it may be due to other factors. One such factor is that the results are a reflection of a Type I error which is when we obtain significant results that should not be significant. It's always a possibility, though the degree of significance is high enough that this would be rare (the likelihood ranges in the .2-1% chance).

Another possibility is that bedsharing is simply serving as a better measure of a third variable (what I was getting at above). In this case, it may better reflect responsiveness in that the willingness to bedshare ever is a willingness to respond to a given child as needed and the needs of individual children to bedshare vary (though biologically most will require close contact during the night at least some of the time), hence the differences in frequency. As the researchers themselves conclude,

"We suggest that there might be greater parental flexibility in parents who, even infrequently, bedshare with their infants. This parental flexibility might translate to daytime practices and less rigid principled parenting. (p9)

Could there actually be a dose-response though?

It is possible that there is a dose-response and this was not accurately captured herein. One possible reason could be the groupings used by the researchers. For example, the group that bedshared 2-3 times per week was not included in the "frequent" bedsharing group even though multiple times per week seems more reflective of frequent bedsharing than a few times per month. Similarly, families who bedshared once a month may be more similar to those who never did it if their children required more contact than they were given. As such, a continuous analysis might better demonstrate a dose response as opposed to the trichotomy created.

A second possibility, in line with what was mentioned above, is that although the research did examine certain elements of child temperament, they did not control for infant sensitivity (instead looking at alertness/responsiveness, unsettled/irregular behaviour, and night awakenings) which may be a greater predictor of the need for constant contact which can be satisfied using bedsharing. That is, it may be that the main effect is driven by a subgroup of children who are more sensitive or "high needs" and this isn't captured in the dose-response categorization.

A third possibility is that the frequent bedsharing group is actually comprised of two distinct subgroups: a planned or responsive bedsharing group and a reactive bedsharing group. Reactive bedsharing refers to cases when bedsharing is done due to pre-existing sleep and behavioural problems and is often not associated with the same outcomes as planned or responsive bedsharing[8]. Notably, it is not necessarily because bedsharing is "bad" but rather that the underlying problem influences outcomes as well. This means that the negative outcomes are most likely linked to the underlying problems, not bedsharing, and bedsharing simply reflects a means of coping with these underlying problems.

What about room-sharing? Was that examined?

This is another factor that may influence the results. In the current study, room-sharing would have been included in the solitary sleep group (due to the nature of the original research design) which we know to be counter to the evolutionary hypotheses about how bedsharing serves attachment needs. It is doubtful that room-sharing at 2 months would be associated with an increased risk of insecure attachment later unless the child was in need of constant contact that could not be met by a hand on the body or simply being close to a parent (which is, for the record, entirely possible for some children; I know as I raised one). However, without the data on where solitary sleepers were sleeping, it is impossible to ascertain the effects of solitary sleeping in another room versus solitary sleeping that occurs proximate to the parents. If, as many would hypothesize, there were no differences in attachment for room-sharing and bed-sharing families, then it would mean the effects for solitary sleeping in their own room not only increase the risk of SIDS[13], but also increase the risk of insecure attachment. This would mean we'd have even more reason to do away with the "nursery" in favour of a family room for the first six months (minimum).

Why is this study missing so many factors?

Whenever research is preliminary, there is usually not a lot of funding for it. This means researchers often piggy back on other research or use pre-existing data sets, which was the case herein. The researchers were able to at least start to look at this question using data from the Generation R Study, but this has inherent limitations, primary of which is that they only have access to the questions that were asked and devised by others. In this case, it left out a lot, including bedsharing patterns over time, room-sharing information, other measures of responsiveness to distress, and so on.

Luckily, the researchers were under no illusion that they were missing these details and they discussed all of these missing variables in their write up, being clear that they are all areas that need to be examined in more detail in future research. They also didn't try to oversell their findings because they are aware that we just have no idea what to make of the lack of dose-response and what the mechanisms would be behind the findings linking solitary sleeping to the increased risk of insecure attachment are. Without more information on that, nothing can really be conclusively stated, though this first finding is critical to getting more research done.

What about people who say "I didn't bedshare and my child is securely attached so the research is meaningless"?

I will repeat here what I have said so many times: People need to understand that increased or decreased risk does not equal certainty. Too often people bring up anecdotes of people that don't "fit" the research and take it to mean the research needs to be thrown out; sadly, this is an effect of a science-illiterate population and one that needs to be addressed. Increased or decreased risk simply refers to the chances and does not dictate individual outcomes. It gives us information about general or population-level outcomes, but does not say a given individual will or will not face that specific effect, only their chances based on patterns in a larger group. In short: use data to help you make decisions, but know that nothing will guarantee any outcome. That's life.