Guidelines from the government and the Royal College of Psychiatrists state that the risk of a patient harming others should be routinely assessed in psychiatric practice (Royal College of Psychiatrists, 1996; Department of Health, 2000). Although this skill was traditionally ascribed to the territory of forensic services, staff in general psychiatry now formally assess risk as frequently as forensic teams (Reference HollowayHolloway, 1997). The increasing expectation on all mental health professionals to identify, appraise and manage risk has led to the introduction of generic risk forms at trust level. Although 67% of trusts have standardised forms for risk assessment, the lack of a gold standard has resulted in striking variation with regards to content, complexity and evidence base (Reference Higgins, Watts and BindmanHiggins et al, 2005).

The Glasgow Risk Screen (GRS) was implemented in January 2004 to support a more transparent, systematic and multidisciplinary approach to clinical risk assessment (NHS Greater Glasgow, 2005). It was developed using the evidence base from an existing and recognised tool (Reference MorganMorgan, 2000; Reference SteinStein, 2005). This bespoke risk assessment tool comprises the categorical identification of several historical, precipitating and protective factors, organised under the four broad risk headings of self-harm, violence, neglect, and other risks. It was determined that the use of the GRS be mandatory at the point of hospital admission and encouraged in other clinical areas.

Previous research has found that the service-wide introduction of a risk assessment tool has not been universally welcomed, with some staff preferring to assess and manage risk by traditional methods and using their own clinical judgement (Reference SteinStein, 2005). There is inter-trust variability in the use of risk assessment forms, and even where mandatory for in-patients the completion rate can be as low as 61% (Reference Higgins, Watts and BindmanHiggins et al, 2005; Reference SteinStein, 2005). To date, there is lack of an adequate evidence base focusing on factors that could influence the completion of generic risk assessment forms.

This study aimed to examine and improve the use of generic risk assessment forms by conducting an audit into the completion of GRS forms.

Methods

The notes for all psychiatric in-patients in south Glasgow were reviewed across two hospital sites on a given day in March 2005. Patients on forensic wards were excluded as they had dedicated forensic risk assessment forms. Data were collected on patient age, gender, admission ward, date and time of admission, detention status, as well as the presence of and adequate completion of a GRS form.

The first intervention was a reminder of the importance of the GRS, sent via email to all clinical staff along with a brief synopsis of the results. The results were also presented at a lecture attended by clinical staff followed by an interactive discussion on risk assessment. Data from the same wards was then re-audited in May 2005 to determine any change in completion rates.

Qualitative feedback received from clinical staff revealed that the GRS forms on some wards were difficult to locate, as were other documentation that required completion during a patient admission. We used a second intervention to address this issue by introducing ‘medical admission packs’ which comprised a plastic folder containing a standardised admission form, physical examination form, prescription chart, patient identifier form and the GRS. Administration staff ensured that all wards had an adequate supply of the packs. The supply of packs to the wards was in itself audited briefly to ensure that availability did not pose a potential confounder. We then re-audited the completion of GRS forms in the same wards in November 2005.

Statistical analyses were performed using version 12.0.1 of SPSS.

Results

At baseline (i.e. initial audit) 109 out of 181 (60%) GRS forms had been completed, 46 (25%) had not been done, and 26 (15%) were incomplete or had problems. After the first intervention of education via lecture and email, 153 out of 189 GRS forms had been completed (81%); 34 (18%) had not been done and 2 (1%) were incomplete. After the second intervention (introduction of medical admission packs) 150 (82%) forms were completed, 33 (17%) were not done and 1 (1%) was incomplete or had problems.

Using χ² tests, we found no statistically significant associations with rates of completion of the GRS forms and (a) whether the form was completed during working hours or out of hours (P=0.401), (b) whether the form was completed for a patient of informal or detained status (P=0.135), or (c) the gender of the patient (P=0.276).

Discussion

Mandatory risk assessment tools ensure that issues surrounding risk are considered and documented as part of patient assessment and management. The use of these forms can thus be central to robust risk assessment and management of psychiatric in-patients.

Previous research has suggested that audit and feedback combined with education results in only a small effect on professional practice (Reference Jamtvedt, Young and KristoffersenJamtvedt et al, 2007). In our study educational interventions in the form of an email and lecture informing clinical staff of audit results improved the completion rates of the GRS by 21%. This dramatic improvement may be related to the fact that our feedback was targeted at the specific population who would complete the forms routinely, i.e. senior house officers. Additionally, the format of our educational interventions not only highlighted the importance of completing the GRS forms, but also emphasised the principles behind a robust risk assessment and underlined its value to clinical management.

Recent evidence has shown that the introduction of a standardised admission form, incorporating clinical risk assessment, resulted in significant improvements to the recording of risk assessments (Reference Dinniss, Dawe and CooperDinniss et al, 2006). However, in our study, the introduction of medical admission packs containing the GRS did not further improve completion rates, although these rates were maintained. Order and carry-over effects from the first intervention are potential biases to be taken into account. This result suggests that the non-completion of risk assessment forms may not be related to their accessibility. Alternatively, the completion rate may have reached saturation point, preventing further increase.

Finally, the literature has demonstrated that problems in the adoption of risk assessment tools such as the GRS frequently relate to misgivings from medical staff regarding the tool's validity, as well as difficulties that multidisciplinary teams have in completing the tool (Reference SteinStein, 2005). We conducted a local survey and also revealed that clinical staff believed a lack of training and being short of confidence in carrying out an adequate risk assessment were additional key reasons for not routinely completing a formulation of risk (NHS Greater Glasgow, 2005). Brief intermittent educational intervention involving clinical staff as employed in our study may thus be helpful in increasing the awareness, and hence confidence, in the use of these risk assessment tools.

Declaration of interest

None.