Are file review-based SAVRY ratings of violence risk reliable?

Since its publication a decade ago, the Structured Assessment for Violence Risk in Youth (SAVRY) has gained acceptance as a strong predictor of future violence in adolescent populations. Clinicians scoring the SAVRY use their professional judgment to code a structured protocol of risk and protective factors based on clinical interviews, a review of the juvenile's records, and other sources of information. Much of the SAVRY validation research, however, has relied upon retrospective ratings obtained solely through file review. To date, no study has examined the reliability of file review-based SAVRY ratings. This study examined whether file-only SAVRY ratings are comparable to expert clinical ratings obtained through standard SAVRY administration procedures. Results indicate that file-only raters were unable to provide Summary Risk Ratings for 43% of the files and were unable to rate 53% of the SAVRY's individual items. The ratings that were coded by the file-only raters had low to moderate levels of agreement with the expert ratings. These results suggest that file-only SAVRY coding is not a reliable manner in which to obtain risk assessment ratings, but the current findings conflict with low rates of missing data in previous file-only SAVRY research. Further research should therefore be undertaken to provide greater clarity as to whether file-only SAVRY ratings of violence risk are reliable.