At the beginning of the QAMC project, it was proposed that the following 20 adverse pregnancy outcomes should be modelled:
Each of these outcomes is an important measure of the quality of care provided to mothers and babies. But, while it is certainly desirable to reduce the incidence of any of these APOs, it does not necessarily make sense to do this by predicting the risk of a mother or baby experiencing such an outcome. For example, it would be of little use to forecast a high risk of maternal re-admission within 14 days of discharge without identifying the root cause of that outcome. This, and other factors, allowed the list of 20 APOs to be reduced to a smaller number of outcomes that are both meaningful and feasible to predict.
Certain items in this list are rendered infeasible to model accurately by their low incidence ( e.g. , maternal death), their absence in any of the perinatal databases we had access to ( e.g. , unintended injury to a maternal organ) or their inability to be precisely defined in terms of database information that we did have access to ( e.g. , postpartum return to delivery suite/theatre).
Further difficulties were caused by some APOs that actually specified a combination of adverse outcomes. Primary Caesarian section for failure to progress is a clear example of this. It is well known that Caesarian section rates vary significantly between different countries, different hospitals and different obstetricians. Any system that tried to predict risk of Caesarian section would have to model the decision process of different obstetricians - a challenging prospect, to say the least. It was decided that these compound APOs should be broken into their component outcomes.
So, after a careful review, the list of 20 APOs was reduced to 5 APOs that seemed feasible to tackle:
Put simply, a failure to progress means that the natural course of labour has stalled before delivery, placing the wellbeing of both mother and child at risk. In terms of the SMR2 data, failure to progress was indicated by the presence of any of the following ICD-9 codes in a case record: