Developing a model for deprescribing in hospital – project detail


  • Develop a model underpinned by empirical evidence and health psychology theory to support practitioners to deprescribe medicines for older people during hospital admission


  1. Characterise deprescribing practice in secondary care
  2. Describe and understand the barriers and enablers of secondary care practitioners to deprescribing for older people during hospital admission
  3. Describe the attitudes of older people and their informal carers towards deprescribing during hospital admission
  4. Develop a theoretically underpinned deprescribing intervention for implementation on an Older People’s Medicine ward


Phase 1 Characterising current deprescribing practice in secondary care

1.1 Audit

Data have already been collected regarding the prevalence of deprescribing in hospital indicating that deprescribing is undertaken most commonly in older people admitted to admissions and Older People’s Medicine (OPM) wards. The medicines most frequently deprescribed in secondary care are opioids, proton pump inhibitors and loop diuretics. The selected reason codes for deprescribing excluding correction of errors were wide ranging. Some were clearly reactive such as ‘suspected toxicity’ whilst others were less clear such as ‘no longer clinically necessary’. An estimate of the ratio of reactive to proactive deprescribing is reactive or proactive will be obtained by reviewing a sample of the notes.

Estimating magnitude of reactive versus proactive deprescribing (6 weeks)

A stratified random sample of 200 notes will be reviewed and text relating to the deprescribing activity will be recorded. Stratification will be by reason code with a greater proportion sampled from reason codes with less certainty regarding whether it was proactive or reactive.

Descriptive statistics will be used to report the proportion of reactive to proactive deprescribing.  If appropriate, inferential statistics will be used to investigate the relationship between medicine and whether deprescribing is reactive or proactive.

1.2 Practitioner perspective focus groups (4 months)

The deprescribing prevalence data and likelihood reactive:proactive categorisation from audit results will be presented to doctors and pharmacists for interpretation in focus groups.

Up to four uni-professional focus groups of eight to ten participants per group will be convened at the hospital. The audit data indicates nine medical specialities are responsible for 467 (79 %) of all deprescribing activity. Three focus groups will comprise senior doctors selected to represent a wide range of the nine specialities and deprescribing activity (determined from the audit data). One focus group will comprise a mix of prescribing and non-prescribing pharmacists of various grades.

Clinical leads from the medical specialities and pharmacy have been notified of the prospective study and expressed interest in their respective specialities being represented in this Trust funded research. The clinical leads will act as gatekeepers and invite appropriate practitioners to the focus groups. By providing them with a hard copy or electronic participant information leaflet and consent form which will include a request for information about suitable dates. Implied consent will be assumed from provision of dates and written, informed consent will be obtained prior to participation in the focus group.

A topic guide informed by the audit data and the theoretical domains framework (TDF) will be developed for use in the practitioner focus groups to:

  • Check agreement with the research team’s categorisation of reactive and proactive deprescribing by presenting a sample of the text recorded from the notes
  • Understand practitioners’ interpretation of the observed frequencies of the different medicines deprescribed
  • Elucidate barriers and enablers to changing practice

The focus groups will be audio recorded and moderated by the applicant. Verbatim transcription will be undertaken by the applicant and checked for accuracy by researcher. A concurrent thematic analysis approach will be used after each focus group. Once thematic analysis has been performed on the transcripts the resultant themes will be mapped to the TDF to identify areas of discussion that require further focus. Thematic analysis will enable coding and analysis to take place without restricting resultant themes to the pre-determined domains of the TDF. A wider understanding of attitudes to deprescribing will be obtained that can then be mapped to the TDF to ensure all targets for behaviour change interventions have been elicited.

1.3 Patient/caregiver perspective questionnaires (4 months)

Ethical and rPATD author approval have been obtained for adapting the patient and caregiver versions of the rPATD using think aloud methods. Adaptation for a UK population will be complete by December 2017.

All eligible patients and caregivers across four OPM wards at a large UK teaching hospital will be invited to complete the rPATD questionnaire. Patients fulfilling the following criteria and people with informal caregiver experience with managing the medicines of people fulfilling the following criteria will be included:

  • Age ≥65 years
  • Taking ≥5 prescribed medicines
  • Inpatient on an OPM ward

Patients and informal carers fulfilling the following criteria will be excluded:

  • Unable to speak and read English
  • Patients are identified by clinical staff as being inappropriate such as being seriously acutely unwell
  • Unable to provide informed consent (identified by ward staff)
  • Patient not under the care of the OPM speciality (applies to patient and carergiver)

To aid with data collection and in accordance with Trust policy on data protection, the research team will administer the rPATD electronically to participants using a mobile tablet device. The research team only will handle the devices by entering responses selected by participants to comply with infection control regulations. Paper copies of the questionnaires will also be made available for participants if requested.

The applicant will be introduced as a researcher to prospective medicine taker participants by a senior doctor during ward rounds. The senior doctor will also provide a very brief account of the proposed research.

After completion of the ward round, the applicant will approach inpatients who expressed interest in study participation. Written informed consent will be obtained prior to administration of the rPATD immediately or at a mutually agreed later time during the study period.

Caregivers will be notified of the research by ward staff and invited to approach the researcher who will be based in a non-clinical section of the ward during visiting hours. Caregivers wishing to participate will provide written informed consent prior to completing the rPATD immediately or at a mutually agreed later time during the study period.

Sample size

Each of the four OPM wards contains approximately 36 beds, thus there are a maximum of 144 occupied OPM beds at one time. The average duration of an admission to hospital for patients aged ≥65 years is between seven and 11 days. Assuming average admission duration of 9 days and three admission cycles per 28 days, there will be 432 patients occupying OPM beds and thus available for recruitment per 28 days.

No participant data are reported for the rPATD, however medicine taker responses to the original PATD indicate that the maximum distribution of participant responses across the response options is 65%:35%. Based on this distribution a sample size of 75 participants will provide a 95% confidence interval of 11%.


An estimated 35% of potential participants will be ineligible:

  • 20% of the older people population will live with a degree of cognitive impairment and are therefore ineligible for inclusion as patients
  • 15% may not fulfil eligibility criteria for reasons such as being too unwell to approach.

A conservative consent rate of 30% in the remaining the 280 eligible participants will provide 84 patients within 28 days.


The proportion of patients receiving informal carer support is unknown; clinicians advise an estimated 50% of patients with cognitive impairment, representing 43 informal carers per 28 days. Assuming that 50% of these carers present during the visiting hours and consent; a sample size of 21 caregiver participants are anticipated. It is therefore envisaged that recruitment of care givers will be completed within four months.

Descriptive statistics will be used to report the magnitude of acceptability of older people and their informal carers towards deprescribing during a hospital admission. Regression analysis will be performed to identify which of the four factors influencing deprescribing contribute most to the overall attitude towards deprescribing.

1.4 Discharge prescription screening

Missed opportunities for deprescribing will be identified by the discharge prescriptions of patients recruited into Phase 1.3 being screened for inappropriate medicines using a validated screening tool: Screening Tool of Older Persons’ potentially inappropriate Prescriptions (STOPP). The methodological approach used for screening discharge prescriptions with STOPP described by Onatade and Auyeyng15 will be adopted.

The number (%) of potentially inappropriate medicines according to STOPP will be reported.  The proportion of patients with one or more PIMs who were amenable to discontinuation based on their rPATD score will also be calculated.

Phase 2 Designing a model for Deprescribing In Secondary Care (DISC)

The quantitative and qualitative data from Phase 1 will inform intervention design. The theory underpinning this process will be the Theoretical Domains Framework (TDF) which will guide selection of appropriate Behaviour Change Techniques (BCTs) to form the intervention components.


Using the TDF and taxonomy of BCTs intervention functions and corresponding BCTs most likely to modify the capabilities, opportunities and motivations of healthcare practitioners (intervention targets) identified in Phase 1.2 will be determined.

The research team will translate the list of intervention targets and BCTs into intervention modules and activities.