Statistics from Altmetric.com
In 1988, I was first asked to give evidence in court in a case of suspected child abuse. I can recall that at that time there was little published research on which to base an opinion. There were certainly plenty of ex cathedra statements in textbooks but all too often this orthodoxy was not founded on empirical science.1 Over 25 years later, it is gratifying to see the extent to which research on non-accidental injury has come out of the shadows. The members of the research team at Cardiff University have been leaders in this field.
The developments of the last quarter century read like a textbook for medical students on different methodological approaches to a research question. Initially, there were case reports and small, single centre case series.2 Then the important cross-sectional study from Montreal.3 This was followed by case control studies.4 Now we have a longitudinal study.5
The authors criticise other studies for using selected populations, such as children attending baby clinics and outpatient appointments. However, this study also recruited children from selected well-baby clinics, hospital outpatient clinics and mother and baby groups. So it was not a geographical cohort study. Nor were the children representative of the population from which they were drawn—46% of the children were from the least deprived quintile and only 15% from the two most deprived …