Disruption of a primary health care domestic violence and abuse service in two London boroughs: interrupted time series evaluation

Background: Domestic violence and abuse (DVA) is experienced by about 1/3 of women globally and remains a major health concern worldwide. IRIS (Identification and Referral to Improve Safety of women affected by DVA) is a complex, system-level, training and support programme, designed to improve the primary healthcare response to DVA. Following a successful trial in England, since 2011 IRIS has been implemented in eleven London boroughs. In two boroughs the service was disrupted temporarily. This study evaluates the impact of that service disruption. Methods: We used anonymised data on daily referrals received by DVA service providers from general practices in two IRIS implementation boroughs that had service disruption for a period of time (six and three months). In line with previous work we refer to these as boroughs B and C. The primary outcome was the number of daily referrals received by the DVA service provider across each borough over 48 months (March 2013-April 2017) in borough B and 42 months (October 2013-April 2017) in borough C. The data were analysed using interrupted-time series, non-linear regression with sensitivity analyses exploring different regression models. Incidence Rate Ratio (IRR), 95% confidence intervals and p-values associated with the disruption were reported for each borough. RESULTS: A mixed-effects negative binomial regression was the best fit model to the data. In borough B, the disruption, lasted for about six months, reducing the referral rate significantly (p = 0.006) by about 70% (95%CI = (23,87%)). In borough C, the three-month service disruption, also significantly (p = 0.005), reduced the referral rate by about 49% (95% CI = (18,68%)). Conclusions: Disrupting the IRIS service substantially reduced the rate of referrals to DVA service providers. Our findings are evidence in favour of continuous funding and staffing of IRIS as a system level programme.