What is the study's methodology?
The team measured the quality of clinical care for insomnia and the rate of adoption and spread of changes in practice using a range of quantitative and qualitative measures.
Prescribing quality indicators were collected and analysed monthly using the following routine Prescribing Analysis and Cost (PACT) data for primary care:
- ‘Average daily quantity’ (ADQ) per Specific Therapeutic Group Age-Sex Weightings Related Prescribing Units (STAR-PUs) for hypnotic drug groups (hypnotic benzodiazepines and Z-drugs) and also individual benzodiazepine and Z-drug hypnotics across participating practices and practices in geographical areas based on commissioning clusters. We planned to plotted and analysed baselines rates and comparisons using statistical process control (SPC) techniques. We anticipated monitoring changes over time using SPC techniques and also employed interrupted times series and regression techniques for observing trend over time and at the time of any intervention with techniques used by the team in other evaluative research initially using practices (in East and Southwest Lincolnshire) matched for prescribing and other demographic characteristics (such as training status, dispensing, rurality) for comparison.
- Identification of high-prescribing practices using comparative confidence charts.
- Hospital inpatient and discharge prescribing of hypnotics using hospital prescribing data.
Focus groups were used to explore the experiences of the prescribing primary care professional and the patient involved in consultations for sleep management. Practitioner attitude, experience and behaviour indicators were measured using validated instruments: validated measures of clinician and patient attitudes, behaviour and experience of therapeutic options for insomnia.
Patient measures included:
Sleep quality improvement such as the Insomnia Severity Index(ISI),4 Pittsburgh Sleep Quality Index (PSQI)5, 6 and Sleep Timing Questionnaire(STQ)7 before and one month after treatment.
- Patient attitudes
Cost indicators: cost analysis from a health service perspective.
Rate of clinician and practice engagement, clinical leadership, and spread: rate of adoption of change management techniques and interventions implemented to improve performance
- Clinical engagement and involvement
- Clinician leaders from practices
- Practice engagement
Healthcare practitioner knowledge and attitudes to quality improvement techniques.
The team initially worked with 8 pilot practices, using rapid experimentation (plan, do, study, act) cycles whilst gathering and analysing data from all practices in the county to enable systematic spread and adoption to over 80% (of 106) practices within 3 years.
The team were able to:
Identify high, moderate and low hypnotic prescribing practices using comparative confidence charts.
Work with willing adopters from moderate and high prescribing practices to develop a network of good practice, measurement and improvement tools, opinion leaders and champions for good practice using a rapid cycle of change.
The team designed simple and complex (multiple), targeted and tailored methods to overcome specific barriers to change:
Involving service users and practitioners to design appropriate interventions
Gathering interview and questionnaire data to identify good practice for tackling inappropriate long-term hypnotic prescribing, improvement methods and potential barriers and facilitators.
The team used a range of active methods to effect change such as:
Minimal (GP letter) to more intensive interventions (gradual tapering with or without cognitive behavioural therapy) for long-term users who want to stop but are unable to despite minimal interventions.
Academic detailing or educational outreach methods selected practices.
Mass media to influence patient and practitioner opinion.
Feedback of performance.
Consideration for psychological therapies for insomnia, possibly using computerised cognitive behavioural therapy as part of current national initiatives to improve access to psychological therapies.
The analysis focused on evaluation of the interventions chosen using an objective outcome variable (PACT data) independent of the interventions. The team observed trend over time, and tested for changes in trend at about the time of the interventions undertaken. We expected that changes may be of several types and we anticipated that we might have to look for sharp changes in those practices that took on interventions rapidly and smooth changes in trend for those who introduced them gradually. We also looked at covariates such as practice (prescribing versus dispensing, training versus non training) and practitioner (single handed versus group) characteristics that might have predicted the extent of change. The team took data one year before and one year after the intervention to remove seasonal effects. Sufficient data points (approximately 12 before and 12 after the intervention using month by month PACT data) were taken to ensure reliability. We intended to report descriptive statistics for the sample as a whole (an intention to change principle) but the analysis of the intervention included the extent to which the practice participated as a covariate (similar to an on-treatment analysis).
The main project took place with practices and practitioners in the rural geographical area of Lincolnshire but we aimed to extend successfully evaluated methods and techniques more widely through regional organisations such as the Commissioning Primary Care Trusts and Strategic Health Authorities and more widely through the East Midlands.
We anticipated that the lessons from this improvement project could, if demonstrated to be effective and cost effective, be applied more generally throughout the whole of the United Kingdom through mechanisms such as the Quality and Outcomes Framework.
We planned to use sustainability and normalisation models to ensure that quality improvements are sustained and spread across the county and more widely both nationally and internationally.
To facilitate spread we developed a spread plan that involved preparation, clear aims and an initial spread plan that could be reviewed and refined as the project progressed.
We prepared for spread by engaging
leadership support for the project and piloting changes in key practice
sites to gain practical evidence for change as well as early
communication of the project and underlying reasons for it to
By 18 months we aimed to involve at least 26 practices (25%) in Lincolnshire commissioning clusters and by 2.5 years at least 75% of practices.