Table 1.

Paper quality scoring system on five criteria:

1	Size of the study population of relevance	Range 0–1
2	Power calculation: whether performed or not	Range 0–1
3	Confounding: degree of adjustment for obvious confounders	Range 0, none; 1, age/gender; 2, age/gender and socio-economic indicators
4	Ethnicity definition: whether explicit, accurate and hypothesis based	Range 0, none; 1, a definition; 2, self assigned; 3 rated on basis of self identification and parental origins
5	Deliberate self harm: whether attempts had been made to ensure the episode was an actual self harm	Range 0, none; 1 self report; 2, some attempt at measurement; 3, well defined and measured

The maximum possible score was 10. This schema was piloted to ensure that it discriminated adequately between the papers included. How well a study measured deliberate self harm and how accurately it measured ethnicity were weighted more highly than other criteria because they were considered particularly important for this review