The Alarm Distress Baby Scale(ADBB) is becoming a widely-used measure of ‘infant social/relational withdrawal’. It is, however, an index rather than a scale and the equal weights assigned to its eight five-level Likert scales have no empirical basis. Moreover, the cut-off conventionally applied to decide on further action (5, from worst of 32) seems to have been derived by considering only its sensitivity and specificity, ignoring the crucial role of prevalence in determining predictive values and hence trade-off between the ‘false alarm’ and ‘false reassurance’ rates. Examination of the main empirical papers shows the numbers of false alarms accepted to avoid one false reassurance ranging from 56 to 2.5 across cut-offs from 4 to 8. In this paper we introduce an online support tool relevant to those making both policy and clinical decision on the use of ABDD. Initially we re-present ABDD as a normalised index, the weights of which can be varied to make it preference-sensitive for a particular local jurisdiction or culture. (The research to determine these tariffs is required, and they can be expected to show culture and sub-culture variation.) Potential users can then enter a range of sensitivities and specificities for ADBB and an estimated population prevalence of ‘infant withdrawal’, to establish the effect of various cut-offs (between 4 and 10) on the relative number of True and False Positives and True and False Negatives generated. At the policy level, decision makers need to place a value on the consequences of each of these four states in order to arrive at an optimal decision, taking into account their frequency. However, legally informed consent to any action based on the index - including referral - requires outlining its potential benefits and harms in the individual case, so those administering the test need to be competent in delivering this.