*2.4. Data Extraction*

We developed a data extraction form based on STROBE criteria for observational studies [16]. For each article the following data were extracted: design, objectives, setting, participants, outcomes, summary measures relevant to research question (e.g., absolute numbers of suicides, suicide rates, standardised mortality ratios, other risk estimates), bias and limitations (as a marker for quality), and interpretation. For each paper included in the study, two authors (AP and HS/RF) independently reviewed each paper, rating quality (including an assessment of study and outcome bias) and summarising findings, including all summary measures presented. Each reviewer completed a data extraction form independently, and compared findings, updating entries in a master data extraction form. Again, a third independent reviewer resolved any discrepancies. As a team we discussed the potential for publication bias, and the potential for selective reporting within studies.

Due to the mixture of summary measures presented, the differing age-groupings, and the discontinuous time periods covered, we used a narrative approach to synthesise temporal findings, taking into account study quality. We used the PRISMA statement to structure findings [17].
