The impact of a behavioural change approach to the implementation of national patient safety guidelines 2014 AIHI Research Symposium Dr Natalie Taylor Prof. Rebecca Lawton Beverley Slater Prof. John Wright Prof. Mohammed Mohammed
A bit about me... University of Leeds Institute of Psychological sciences Bradford Institute for Health Research Australian Institute of Health Innovation
Introduction Project aim Support the implementation of patient safety alerts across NHS hospitals Funding Regional Innovation Fund (RIF) Yorkshire & Humber Health Innovation and Education Cluster Today s presentation Background (patient safety, theory, implementation) Methods Interventions Results Limitations and future research Translation and sustainability
START T O P D O W N
Why is it difficult for health care organisations? Ward environments are busy, chaotic Medical staff autonomous All hospitals are different Alert guidelines lack transparency Alerts release rates overwhelming Top down approach NPSA to be abolished Management need to tick boxes Trusts can self-declare compliance Habits Changing behaviour is difficult and complex Methods used often lack the necessary components that are effective in producing change
Example: Reducing the harm caused by misplaced nasogastric feeding tubes Summary Problem (X-ray misinterpretation) Behaviour change gap Recommend action (17 guidelines; one = check ph first line) No guidance on how to ensure staff perform recommended actions Behaviour change interventions?
Benefits of theoretically informed interventions Is theory important for intervention design? Explain the structural and psychological determinants of behaviour Allows for more accurate replication of interventions Guide the refinement of health promotion interventions How can interventions can be informed by theory? Provides theoretical constructs to target (e.g., barriers) Able to target constructs with evidence based behaviour change techniques Allows for identification of mechanisms of effective interventions Claims for increased effectiveness of theory based interventions Date back to the 1980 s, E.g., Green (1984); Grffin et al. (1999); Marteau et al. (2006); Michie et al. (2007); Painter et al. (2008) Meta-analyses of interventions for health behaviour change (Albarracin et al., 2005; Taylor et al., 2012)
Is it possible to use theory-based interventions in complex healthcare settings? Perhaps...if you combine with principles of implementation: Management approval and ongoing support Commitment amongst members of the target group Use of boundary spanners Mapping of guidelines onto local problems Adopting the perspective of the target group Acknowledging the complexity of the changing behaviour in practice A monitoring plan A flexible approach that is driven by local context Co-production and design to combine theoretical and contextual expertise Incorporation into established structures
Overview of methods Four NHS Trusts across Yorkshire and the Humber Six sites in the north of UK Two NPSA alerts per Trust Reducing the risk of overdose of midazolam injection in adults x 2 Promoting safer use of injectable medicines x 1 Medicines reconciliation x 1 Reducing harm caused by misplaced nasogastric feeding tubes x 3 Retrospective control data collected
The Theoretical Domains Framework Implementation (TDFI) approach (Taylor et al., 2013a,b) Stepped process informed by behaviour change theory and implementation literature (Michie et al., 2005, 2008; Grol et al., 2007) Healthcare professionals not using ph as the first line method for checking tube position Involve stakeholders Medical directors and sharp end staff Identify target behaviour Audit and discussion Identify barriers Influences on Patient Safety Behaviours Questionnaire (IPSBQ) Confirm barriers and generate intervention strategies Focus groups Support staff to implement and evaluate intervention Joint approach Re-auditing
Questionnaire results Barrier Barriers to using ph as the first line method for checking tube position Mean (SD) H1 n = 99 Mean (SD) H2 n =105 Mean (SD) H3 n =23 Mean (SD) all hospitals n = 227 Inter-item correlation Knowledge 2.02 (0.70) 2.33 (0.75) 2.08 (0.76) 2.17 (0.74)** 0.64 Skills 2.37 (0.79) 2.64 (0.72) 2.74 (0.87) 2.53 (0.78)** 0.62 Social and professional identity 2.04 (0.73) 1.96 (0.64) 2.16 (0.79) 2.01 (0.69) 0.23 Beliefs about capabilities 2.44 (0.77) 2.55 (0.83) 2.52 (0.97) 2.50 (0.81) 0.43 Beliefs about consequences 2.35 (0.70) 2.38 (0.70) 2.39 (0.48) 2.37 (0.68) 0.45 Motivation and goals 2.40 (0.66) 2.40 (0.60) 2.65 (0.69) 2.42 (0.64) 0.21 Cognitive processes, memory and decision making Environmental context and resources 2.36 (0.68) 2.47 (0.74) 2.19 (0.67) 2.39 (0.71) 0.23 2.55 (0.85) 2.69 (0.69) 2.68 (0.62 2.63 (0.76) 0.47 Social influences 2.84 (0.76) 2.89 (0.73) 2.71 (0.75) 2.85 (0.74) 0.22 Emotion 2.41 (0.65) 2.75 (0.55) 2.35 (0.62) 2.56 (0.63)* 0.62 Action Planning 2.32 (0.66) 2.38 (0.62) 2.42 (0.54) 2.36 (0.63) 0.43
Focus group results: interventions matched to barriers and BCTs (H1) Barrier Strategy Behaviour change technique* Social influences Emotion Environmental context and resources Bcap (and knowledge and skills) Information presented at clinical governance meetings by experts in the area Awareness day held within the Trust Posters with pictures of senior staff performing correct behaviour Screensaver contained messages to elicit anticipated regret and to reframe perspective on behaviour Radiology and ward protocols to empower staff Instructions, flow chart, measurement tool, who placed NG, place to record ph values, etc. Splashscreen placed on intranet with prompt about ph testing and link to all relevant documentation Practical training complete for current FY1s E-learning package developed for junior doctors Persuasive source Information about health consequences, and social/ environmental consequences Prompts, cues, social support (unspecified) Anticipated regret Salience of consequences Framing/reframing Prompts, triggers, cues Adding objects to the environment Instruction on how to perform a behaviour Behavioural practice/rehearsal
Practice change results Target behaviour: Using ph as the first line method for checking tube position Audit information Hospital 1 Hospital 2 Hospital 3 Hospital 4 (Control) Pre Post Pre Post Pre Post Pre Post Number of sets of notes audited 49 48 43 44 44 40 53 46 ph of aspirate from stomach 18% 63% 12% 73% 14% 33% 45% 46% Patient sent for X-ray 49% 23% 77% 9% 41% 40% 25% 20% Tube placed in radiology 0 0 0 0 36% 10% 0 0 Information not documented 33% 15% 9% 18% 9% 18% 30% 46%
Practice change results Zou s modified Poisson regression approach Response variable: ph test used first line (yes/no) Covariates: 1) hospital and 2) an interaction term with a binary pre/post variable (0/1) represented by a combination of the variables x-ray, radiology, ND Coefficients from the model represent risk ratios with respect to the control hospital separately for the pre-intervention and post-intervention time periods Following intervention implementation, the use of ph first line increased significantly across intervention hospitals compared to the control hospital, which remained unchanged First Line method ph Hospital Pre % Post % Pre model coefficients with respect to the control Risk Ratio (95%CI) p-value for model coefficients with respect to the control hospital Post model coefficients with respect to the control hospital Risk Ratio (95%CI) p-value for model coefficients with respect to the control hospital H1 18 63 0.41 (0.21 to 0.79) 0.007 4.43 (1.99 to 9.87) < 0.001 H2 12 73 0.26 (0.11 to 0.62) 0.002 8.14 (3.06 to 21.67) < 0.001 H3 14 33 0.30 (0.14 to 0.67) 0.003 3.1 (1.14 to 8.43) < 0.05 H4 (control) 45 3 Reference 0.77 (0.47 to 1.26) 0.296
% of pateints with NG feeding tubes who had ph testing as the first line test method following insertion 100% 90% 80% Sept & Oct 2011: project presented at 4 clinical audit meetings 70% 60% 50% 40% June 11: new trust NGT documentation February 4th 2012; screen saver launched with an awareness day Junior doctor rotation % total numbers % minus theatre 30% 20% March 2011:revised NPSA alert released 10% October 2011; FY1 doctors attend mandatory NGT training 0%
B O T T O M U P
Limitations and future work Limitations Research design retrospective control group Impact of confounding variables Unable to identify effectiveness of individual interventions, BCTs Future research Test the TDFI approach in Australian hospitals More rigorous research design Different guidelines
Translation and sustainability The TDFI approach Being taken forward by the Yorkshire and Humber Academic Health Science Networks covering a population of 5.3m (10% of England) 1-day workshop and toolkit ABC for Patient Safety Improvement Academy staff trained in the use of the toolkit Three workshops delivered since March 2013; next = June Plan to run workshops four times/year to enable wide access
Thanks to Sally Moore Victoria Robins Lorna Peach Shaan Hyder Sahdia Parveen All our hospitals and steering group members And You for listening Questions welcome
Results: other alerts Midazolam ; Injectable medicines; Medicines reconciliation Hosp Target behaviour Indicators(s) and desired direction of change ( / ) N pre N post Pre-audit results Post-audit results N pre N pos Control pre Control post H1 H3 H4 Titrate doses according to individual patient needs Serum gentamicin levels taken between 6-14 hrs of patient receiving gentamicin Compile accurate inpatient prescriptions (doctors) Effectively communicate any changes, omissions, or discrepancies to doctors (pharmacists) Mean total amount of midazolam (mg) ( ) Mean amount of midazolam (mg) per dose ( ) N (%) blood levels checked between 6-14 hours ( ) Total (SD) drugs omitted ( ) Total (SD) discrepancies ( ) Communicated discrepancies ( ) 30 26 4.5 (3.07) 2.7 (1.62) 63 51 4.4 (3.36) 4.2 (2.31) 3.3 (1.90) 2.2 (1.12) 3.8 (1.71) 3.2 (1.69) 43 22 25/43 (58) 21/22 (95) 43 22 29/43 (67) 19/22 (86) 39 22 1.59 (2.1) 1.10 (1.14) 37 39 1.03 (1.44) 1.28 (1.49) 2.03 (2.5) 1.13 (1.23) 3.32 (5.85) 1.79 (2.39) 11/23 (49) 12/16 (75) 20/27 (74) 13/23 (57)