The PRASE intervention process
The PRASE intervention is designed to collect feedback from patients about the safety of their care, using robust tools developed within our team over the past 3 years. This patient feedback is then collated and presented to each ward in a formal report (a ‘feedback report’). This report then allows ward staff to understand more about how patients perceive the safety of their care on their ward, and then target improvements based on problematic areas. The philosophy of this intervention is that it is an iterative process with a cycle of measurement, feedback and change lasting for a period of 6 months. The three key stages of the process are: a) measurement, b) feedback, c) action planning and change.
a) Measurement
Over a 3- to 4-week period, an average of 25 patients per ward will be recruited to participate in the measurement phase. Each patient (deemed to have capacity to consent) will be approached, the study explained, and informed consent taken. Following this, and using a computer tablet, the research fellow or research nurse will ask the patient to complete the 44-item PMOS questionnaire, and report any safety concerns (or specific positive experiences of care) using the PIRT incident reporting tool. The 44-item questionnaire asks questions based on eight different domains which are:
-
communication and team working
-
organisation and care planning
-
access to resources
-
ward type and layout
-
information flow
-
staff roles and responsibilities
-
equipment
-
delays
An additional four questions are asked, which are described under the Secondary Outcomes section. Patients will be given a choice of whether they would prefer to self-complete the questionnaire or have it facilitated by the researcher. Patient reports of safety concerns (or positive reports) can only be recorded in a facilitated conversation with a researcher, due to the need to elicit detailed information in a specific format that needs to be typed onto the computer tablet. The PMOS questionnaire and PIRT tool are detailed in Additional file
1: Appendix 1 and Additional file
1: Appendix 2.
There are three measurement periods – at baseline, at 6 months, and a final measurement at 12 months.
b) Feedback
Following the measurement period, the information for each ward will be collated and presented to the ward in the form of a ‘Feedback Report’. This report provides a variety of different qualitative and quantitative information for staff, and has been designed and piloted by a team of patients, academics and health professionals to be as user friendly as possible. Please see Additional file
1: Appendix 3 for an example of the feedback report.
The report provides an overall ward safety profile (on pages 3 and 4) which summarises scores and number of reports (concerns or positive experiences) relating to each contributory factors domain of the PMOS questionnaire, as well as providing a breakdown of how these scores and reports relate to specific questions. PMOS questionnaire scores are shown graphically using a traffic light system to allow staff to see where they are performing well, and where improvements are necessary.
No recommendations for areas of action are suggested by the research team. The feedback report is simply a reflection of the patient’s perspective of the safety of their care. It is then up to ward staff to identify areas to target for improvement, within the next phase of the intervention.
c) Action planning and change
The next phase of the intervention is action planning, followed by implementing and monitoring changes, based on the areas identified for action in the feedback report. To undertake the action planning, we will ask participating wards to identify an Action Planning Team (APT). This team will comprise a minimum of four people who work on the ward, and ideally include both senior and more junior staff, from different professional groups. An example action planning team might include any of the following representatives: matron, consultant, ward manager/ward sister, junior doctor, staff nurse, healthcare assistant, Allied Health Professional (for example, a physiotherapist, an occupational therapist), pharmacist, ward clerk or patient representative. The APT will be responsible for receiving the feedback report, considering which area(s) should be targeted, and agreeing an action plan for improvement. In addition, the team will need to monitor the implementation of the plan. A nominated person within the APT will take responsibility for delivering the action plan.
Action Planning Meetings (APMs) will be facilitated by a senior researcher from the Quality & Safety research team. The piloting of the PRASE intervention showed that facilitation of APMs was important for generation of concrete action plans as an outcome of the action planning meeting.
Control wards will receive no intervention during the study duration. At the end of the study, control wards will receive all their feedback reports for the three time points of data collection amalgamated into one report.
Setting and sample
This study will be undertaken within 32 hospital wards, spread across three NHS Trusts, over five different hospital sites (eight wards at a small district general hospital, 10 wards at a medium sized teaching hospital and 14 wards at a very large teaching hospital). Study participants will be patients within participating wards. An average of 25 patients within each ward will be recruited at three different time points across the study period. Each time point lasts 3 to 4 weeks and data collection occurs within the same 3- to 4-week window for all participating wards in the same Trust. The eligibility criteria for the study are given below but, briefly, the selection criteria for participation is: any patient aged 16 years or over who has capacity to give informed consent to take part. All patients who satisfy the eligibility criteria are approached and then if they agree to take part in the study and give informed consent then they are recruited into the study. Therefore, selection bias is minimal as researchers approach every patient on the ward in a stepwise fashion, unless the patient does not have mental capacity to give informed consent or they are so gravely ill or distressed to the extent that it would preclude the researcher undertaking the PMOS questionnaire with them. Recruitment of patients on a ward ceases when 25 patients have been recruited. The study will involve 800 patients at each time point and equates to a total sample of 2,400 patients. While study participants are patients, the outcome measures are at ward-level meaning that the unit of analysis is the ward.
Wards will be randomly assigned to either the intervention or control groups, on a 1:1 ratio. Randomisation will be carried out by York Trials Unit Randomisation Service (based at the University of York) using a secure computer system. All wards from each Trust will be randomised in a batch. Although there is the possibility of staff moving between control and intervention wards in a hospital, the possibility of contamination between the groups is expected to be minimal. This is because the ‘Feedback Report’ is not available to control wards and the intervention is ward-specific dependent on this. Baseline data collection will be undertaken at least 1 month before randomisation to ensure completion of data collection before knowledge of randomisation. Once baseline data have been collected in a Trust, participating wards will be randomly allocated to one of the two arms: intervention or control. Minimisation will be used to balance the groups with respect to ward specialty, average age, single/mixed sex wards and ward size. After randomisation, ‘start up’ meetings will be held with intervention wards only. Intervention wards will be asked to set up action planning meetings in order to consider their first feedback report. The above process is detailed in flow chart format in Additional file
1: Appendix 4.
Trial analysis
A full Statistical Analysis Plan will be written prior to any analyses. Analyses of individual level outcomes will account for clustering. For the primary outcomes, the difference between intervention groups will be estimated with 95% confidence intervals. The intervention groups will be compared with respect to the proportion of harm-free care across wards at 12 months. A linear regression model accounting for the minimisation factors and the baseline level of harm-free care will also be included as a covariate. As the outcome measure is on a ward level, no adjustment for clustering is required. The difference between the intervention groups at 12 months will be calculated with a 95% confidence interval. The standardised effect size will also be reported.
Multi-item PMOS domains will be treated as continuous data in the analysis. Linear mixed models accounting for the minimisation factors and baseline PMOS scores (ward level averages) will be used to compare the intervention groups with respect to each domain score at 12 months. Random effects will be used to account for the clustering at ward level. The difference in adjusted (least square) means will be summarised with 95% confidence intervals. Standardised effect size and ICC will also be calculated. Due to the nature of the intervention it is not possible to identify which domains are of primary interest, since the intervention across different wards may focus on improving different domains. Each domain will therefore be tested without adjustment for multiplicity. Interpretation of the P values will be considered carefully alongside the full details of which domain(s) the interventions in each ward were aiming to improve.
Secondary outcomes will be analysed using similar models to the primary outcomes. Ward level covariates will be added to the primary model in order to explore their influence. Continuous measures will be compared between groups using multi-level regression modelling accounting for trust/ward where required and the minimisation factors. Categorical measures will be compared between groups using logistic regression (or ordinal logistic regression for ordinal measures) adjusting for the Trust/ward where required and minimisation factors. The validity and reliability of the PMOS and domain scores will be investigated in this population using Cronbach’s alpha and known-group comparisons. All primary and secondary outcome measures will be summarised at each time point and by intervention group using summary tables or graphs/bar charts depending on the type of data.