What is the M-ADBB ?

The M-ADBB is the modified version of the ADBB scale (modify-ADBB) which was developed in 2005 by Prof. Stephen Matthey’s team, in collaboration with Prof. Guedeney, creator of the original ADBB in Australia.

This simplification of the ADBB scale was intended to be more adapted to the Australian context for wider use in routine consultations.

What are the differences with the original scale?

Three items were deleted taking into account the results of psychometric studies of the ADBB scale already conducted.

Self Stimulating gesture item (n°4)

Because it seems difficult to obtain a good inter-judge agreement, even with health professionals experienced in pediatrics. The risk of error is then too high.

Furthermore, psychometric studies show that this is an isolated item in several factorial analyses of the ADBB scale.

This item has a very high difficulty parameter, indicating that it is discriminating for very high levels of relational withdrawal. This is not necessary for a tool intended to be used in a universal screening context like the M-ADBB.

Attrativity Item  (n°8)

This item poses many confounding problems for professionals using the ADBB and has a high correlation with six other items based on principal component analysis. This indicates that this item does not provide sufficient information on its own. 

In addition, the attractiveness item revealed a strong dependence of this item with item #7 (“relationship with observer”) suggesting that both perform equally well in differentiating an infant with the same level of relational withdrawal. Thus, it could be removed from the scale without much difficulty.

The item “Responsiveness (or briskness of response)”(n°6)

This item has a strong correlation with the item “general activity level” (n°3) according to the principal component analysis. 

It is also very difficult to evaluate and is only triggered when the level of withdrawal is high, so it is not suitable for a universal screening tool.


When the ADBB was first used in Australia, nurses used to rate each item as “satisfactory”, “possible problem” or “definite problem” in order to proceed more quickly.

This rating system has been retained and can eventually be converted into numerical values:
– satisfactory = 0
– possible problem = 1
– certain problem = 2

There is an exception for the vocalization item:

The vocalizations being sometimes of late appearance at the time of the examination because of a justified fear of the child or the need of time before this one vocalizes (very related to the temperament).

Thus, for this item, only two options are possible: “satisfactory = 0” or “possible problem = 1”.


The M-ADBB is a very good tool to identify babies who will need special attention and a more thorough evaluation for front line professionals.

It can be acquired in 3 hours through an online webinar that can be organized specifically for your team (in-house training) or on specific dates that are updated regularly (inter-house training).


For more information contact Jocelyne Guillon at info@humagogie.fr

Correlation with the ADBB scale

A high correlation (greater than .80) between an instrument whose validity is to be tested and another version of the instrument whose validity has been demonstrated allows both versions to be considered as measuring the same theoretical model as the version whose validity has already been established.

The pilot study calculated the correlation coefficient between the ADBB and M-ADBB. The coefficient obtained was 0.72.
This same study concluded that a score ≥ 2 for the M-ADBB is equivalent to the cutoff score of 5 for the ADBB.

Another study calculated a correlation coefficient between ADBB scores obtained by experts and M-ADBB scores obtained by trained professionals: Correlation coefficient = 0.82 CI [0.68-0.91] (p < 0.001).

These results demonstrate the relevance of the M-ADBB.


It is about demonstrating that the same tool used by different people detects the same babies.

Inter-judge reliability: The authors of the M-ADBB scale obtained excellent inter-judge agreement: coefficient ranging from 0.87 to 0.90.

The agreement was also good in the Nepalese study: ICC = 0.81 (0.71-0.91). And this result was significantly better than for ADBB (ICC 0.39) in the same team.

 Therefor the ADBB scale is a concensus tool.