BackgroundDomestic violence and abuse remains a major health concern. It is unknown whether the improved healthcare response to domestic violence and abuse demonstrated in a cluster randomised controlled trial of IRIS (Identification and Referral to Improve Safety), a complex intervention, including general practice based training, support and referral programme, can be achieved outside a trial setting. Aim: To evaluate the impact over four years of a system wide implementation of IRIS, sequentially into multiple areas, outside the setting of a trial.MethodsAn interrupted time series analysis of referrals received by domestic violence and abuse workers from 201 general practices, in five northeast London boroughs; alongside a mixed methods process evaluation and qualitative analysis. Segmented regression interrupted time series analysis to estimate impact of the IRIS intervention over a 53-month period. A secondary analysis compares the segmented regression analysis in each of the four implementation boroughs, with a fifth comparator borough.DiscussionThis is the first interrupted time series analysis of an intervention to improve the health care response to domestic violence. The findings will characterise the impact of IRIS implementation outside a trial setting and its suitability for national implementation in the United Kingdom.Electronic supplementary materialThe online version of this article (10.1186/s12889-018-5865-z) contains supplementary material, which is available to authorized users.