Abstract
Background:
There is no conclusive evidence that either high doses or combinations of antipsychotics are more effective than standard doses or monotherapy alone. Nonetheless, prescription of both remains prevalent in the UK. In 2006 the South London and Maudsley NHS Foundation Trust (SLAM) participated in a national survey of prescription of antipsychotic medications, organized by the Prescribing Observatory for Mental Health. Over half of the patients on SLAM inpatient or psychiatric intensive care units were prescribed a high-dose antipsychotic or a combination of antipsychotics. Prescribing high-dose antipsychotics and polypharmacy in SLAM was found to be among the highest in the UK.
Aim:
To assess the impact of a 6-year quality improvement programme aimed at reducing the rates of prescribing high-dose antipsychotics and polypharmacy on SLAM inpatients and psychiatric intensive care units.
Results:
There was a significant reduction between baseline and final survey in the rates of prescription of both high-dose antipsychotics and polypharmacy on SLAM inpatients and intensive care units (58% versus 10% p < 0.0001 and 57% versus 16%, p < 0.0001 respectively). The proportion of patients at final survey prescribed a high-dose antipsychotic and a combination was substantially lower in SLAM than in the national sample (10% versus 28%, p < 0.0001 and 16% versus 38%, p < 0.0001 respectively).
Clinical implications:
A sustained change in the prescribing culture of an organization can be achieved through a targeted improvement programme.
Introduction
Antipsychotic drugs are effective in the acute and long-term management of schizophrenia, psychosis and bipolar disorder. Despite the widespread practice of antipsychotic polypharmacy [Paton et al. 2008; Prescribing Observatory for Mental Health, 2012] there is no firm evidence that high doses or combinations of antipsychotics are more effective than standard doses or monotherapy. Similarly, the use of ‘as required’ antipsychotics, though common in the UK, is poorly supported by available data [Barnes et al. 2011]. Furthermore, the addition of an antipsychotic for ‘as required’ use has been shown to be a contributory factor in those prescribed a high-dose antipsychotic [Harrington et al. 2002; Paton et al. 2008; Milton et al. 1998]. The majority of antipsychotic adverse effects, including extrapyramidal effects, hyperprolactinaemia, QTc prolongation, sedation and anticholinergic effects are dose related. The use of high-dose antipsychotics is associated with an increased incidence and severity of these adverse effects [Centorrino et al. 2004].
In 2006 the Royal College of Psychiatrists issued a consensus statement in which it recommended that the routine use of high-dose antipsychotics in adult general mental health services should be avoided and advised that the decision to prescribe a high dose only be taken once evidence-based strategies have been exhausted [Royal College of Psychiatrists, 2006]. The UK National Institute for Health and Care Excellence guidelines for schizophrenia which followed shortly after recommended that antipsychotic doses be prescribed within the British National Formulary dose range and that combinations of antipsychotics for regular use be avoided [National Institute for Health and Care Excellence, 2014].
The recommendations appear to have done little to change practice and the prescription of both high-dose antipsychotics and polypharmacy remains highly prevalent in the UK [Harrington et al. 2002; Mace and Taylor, 2005; Paton et al. 2008; Prescribing Observatory for Mental Health, 2012].
Programmes aimed at reducing antipsychotic polypharmacy have produced only very little change [Milton et al. 1998; Paton et al. 2008; Thompson et al. 2008]. In this paper we report the results of a baseline survey and the impact of a subsequent 6-year quality improvement programme aimed at reducing the rates of high-dose antipsychotics and polypharmacy in the South London and Maudsley NHS Foundation Trust (SLAM).
Method
This programme was carried out in SLAM, an inner city trust which serves a multicultural population of approximately 1.2 million with high rates of immigration and social deprivation
The audit was performed in several stages: the baseline survey, followed by two further surveys, each followed by a quality improvement programme and the final survey.
Ethics approval was not sought for this audit as it was considered within the trust’s commitment to Prescribing Observatory of Mental Health (POMH-UK) programme. Data were seen and manipulated only by clinical staff directly involved in patient care. Data were anonymized and stored on a computer which was password protected.
Definitions
High dose was defined as a total daily dose (whether of a single antipsychotic or combined antipsychotics) greater than 100% of the maximum dose recommended in the European Public Assessment Report (EPAR), published by the European Medicines Agency for that drug.
Antipsychotic polypharmacy was defined as the prescription of two or more antipsychotic drugs (either regular or ‘as required’). The total daily dose was calculated as the sum of the individual antipsychotic drugs when expressed as a percentage of the EPAR maximum dose. Inclusion criteria were all patients prescribed an antipsychotic on SLAM inpatient and psychiatric intensive care units (PICUs).
Stage 1: baseline (first) survey and feedback of results
In January 2006 the trust participated in a national survey of prescription of high-dose antipsychotics and combinations of antipsychotics. All patients on an inpatient or PICU who were prescribed an antipsychotic were identified. Names and doses of any regular and ‘as required’ antipsychotics were noted from prescription charts. For ‘as required’ antipsychotics the maximum prescribed dose for the previous 24 h was recorded. Data were submitted electronically to the POMH-UK. The results were disseminated in the trust through the Drug and Therapeutics, Medicines Management and the Formal Executive Committees. The results were also disseminated locally through smaller clinical groups.
Stage 2: the first quality improvement programme
Following the baseline survey a quality improvement programme was designed by the trust pharmacy department for implementation in all services across the trust. The following interventions were introduced.
March–October 2006
Pharmacy agreed with trust clinicians the following restrictions on the use of ‘as required’ medications. First, antipsychotics for ‘as required’ use were not to be routinely prescribed. Second, that all ‘as required’ antipsychotic prescriptions were to be reviewed at least once a week. These restrictions and guidance on prescribing ‘as required’ antipsychotics were approved by the Trust Executive and disseminated throughout the trust. Prescribers were reminded to individualize any prescription for an ‘as required’ antipsychotic and to specify for each ‘as required’ prescription the dose (including maximum dose), frequency and single route of administration, as well as the precise circumstances under which the drug may be administered.
April 2006–January 2007
Pharmacy identified all inpatient prescriptions for high doses and combinations of antipsychotics. Pharmacists, with prescribers, reviewed these prescriptions and if possible alternative treatment options were introduced. When a prescription was deemed necessary, reasons for continuation and a plan for any additional monitoring were recorded in the patient’s notes.
Restrictions and guidance (as outlined above) on the use of ‘as required’ medications were implemented on inpatient units.
Stage 3: the second survey
Data for the second survey were collected in January 2007, as detailed in stage 1.
Stage 4: the second quality improvement programme
The quality improvement programme outlined in stage 2 was repeated between March and December 2007. Rates of prescribing high-dose antipsychotics and combinations were compared across the trust. Practice was compared for wards with a similar patient demographic (for example, female acute units from different boroughs or clinical specialties, such as PICUS). Results were reported to the relevant staff.
Stage 5: the third survey
Data for the third survey were collected in January 2009, as detailed in stage 1.
Stage 6: the third quality improvement programme
The quality improvement programme outlined in stage 4 was repeated between March and December 2009. In addition, the following interventions were introduced.
January 2009
Pharmacy, through the trust’s Executive Performance Management Review process, agreed a target with all trust services to reduce the rates of prescribing high doses and combinations of antipsychotics on individual units to below 20% by the end of 2009.
February 2009–December 2011
Prescriptions were examined on units with disproportionately high rates of prescription of either high doses or combinations of antipsychotics.
April 2009
Trust inpatient prescriptions were updated to include a warning that all ‘as required’ medications must be reviewed at least once a week.
Stage 7: final survey
Data for the final survey were collected in January 2012, as detailed in stage 1.
Data analysis
Categorical variables were presented as percentages. χ2 tests were used to assess the difference in the rates of high-dose and combination antipsychotics in SLAM between baseline and final surveys, and the difference in rates of high-dose and combination antipsychotics between SLAM and the national sample at final survey. A value of p less than 0.05 was considered statistically significant.
Results
More than 200 patients were included in each of our surveys. There was a significant reduction in the rates of prescribing high-dose antipsychotics (58% versus 10%, p < 0.0001) and polypharmacy (57% versus 16%, p < 0.0001) between baseline and final survey (Table 1).
A summary of results (baseline, second, third and final surveys).
SLAM, South London and Maudsley NHS Foundation Trust.
The proportion of patients at final survey prescribed a high-dose antipsychotic and a combination of antipsychotics was significantly lower in SLAM than in the national sample (10% versus 28%, p < 0.0001 and 16% versus 38%, p < 0.0001 respectively) (Table 2 and figures 1 and 2).
Proportion of patients prescribed an antipsychotic high-dose and combination at final survey in SLAM and the national sample.
SLAM, South London and Maudsley NHS Foundation Trust.
Proportion of patients in South London and Maudsley NHS Foundation Trust (SLAM) and the UK prescribed a high-dose antipsychotic in each survey.
Proportion of patients in South London and Maudsley NHS Foundation Trust (SLAM) and the UK prescribed an antipsychotic combination in each survey.
Discussion
Our results show a significant reduction in the rates of prescribing high doses and combinations of antipsychotics on SLAM inpatient and PICUs. Improvement was noted after each stage and continued over the 6-year duration of the programme. Rates of high-dose antipsychotics and polypharmacy in SLAM are now substantially lower than in the rest of the UK.
Our baseline survey in 2006 revealed a high rate of ‘as required’ antipsychotic prescribing; over half of all SLAM patients were prescribed a combination of antipsychotics, two-thirds of which were indicated for the control of acutely disturbed behaviour. Antipsychotic ‘as required’ prescriptions accounted for the majority of the antipsychotic polypharmacy. The practice of ‘as required’ prescribing was noted to be routine and commonplace across the trust for patients on admission. Furthermore, prescription often continued throughout the admission despite patients being stabilized on regular antipsychotic medication.
Prescribing ‘as required’ antipsychotics on psychiatric inpatient units in the UK is common [Harrington et al. 2002; Paton et al. 2008; Prescribing Observatory for Mental Health, 2012] despite the lack of evidence to support its use [Whicher et al. 2002]. The decision to prescribe is a medical one but administration of ‘as required’ drugs is at the discretion of nursing staff [Harrington et al. 2002]. For the purpose of our audit, for ‘as required’, the maximum dose prescribed and able to be administered by nursing staff within the previous 24 h was noted as the prescription. This was viewed by many prescribers in the trust as an unfair reflection of practice; that is, ‘as required’ medication prescribed is often not given. Perceptions of the efficacy of antipsychotic medications have been shown to differ between nurses and doctors. In one questionnaire survey, more nurses than junior doctors preferred an antipsychotic to a benzodiazepine for the control of agitation and acute psychosis [Geffen et al. 2002]. Certainly, from our discussions in the trust it was evident that staff on acute inpatient units welcomed the option of an antipsychotic for administration when necessary. ‘As required’ medications were not routinely reviewed on wards, or generally viewed as part of a patient’s overall drug burden.
Our results are consistent with those of a national survey which showed that ‘as required’ prescribing was a major contributory factor in high-dose and combination prescribing [Paton et al. 2008]. There is no conclusive evidence that high doses of antipsychotics are more effective than standard doses. Dose–response relationships for antipsychotics tend to show a threshold-type effect and optimum doses appear to be well within the licensed dose range [Ezewuzie and Taylor, 2006; Sparshatt et al. 2008; Kinon et al. 2008; Mace and Taylor, 2009]. The risks of high-dose prescribing are well known [Baldessarini et al. 1988]. Nonetheless, in practice a belief amongst clinicians about the benefits of high doses appears to persist. Nursing staff caring for patients prescribed a high antipsychotic dose, when asked in a survey, agreed with the need for higher doses or antipsychotic combinations for symptom control. Interestingly, in the same survey, psychiatrists prescribing a high dose were sceptical of the validity of evidence supporting prescribing algorithms, and did not feel prescribing algorithms were necessary in clinical practice [Ito et al. 2005]. In practice, doses are often increased when patients inadequately respond to treatment or where a more immediate response is required [Barnes and Paton, 2011].
Our first intervention, which was to agree and implement restrictions on the prescription of medication for ‘as required’ use, led to a reduction in the overall rate of prescribing high doses and combinations of antipsychotics. Fewer patients at reaudit were noted to be prescribed antipsychotic polypharmacy for the management of acutely disturbed behaviour. We believe this intervention was successful because discussions about the need for change were conducted widely throughout the trust, with staff of all grades. Through these meetings the appropriateness of the existing culture of routine ‘as required’ use was challenged. In addition, the plan was endorsed by the trust executive committees, which provided the organizational authority helpful in delivering change.
The other indication commonly recorded in patients’ notes for the use of antipsychotic combinations was a poor response to monotherapy. ‘Augmentation’ due to poor response was not limited to clozapine. Evidence of the effectiveness of combined antipsychotics is scant [Lochmann van Bennekom et al. 2013]. This is also true for the now largely accepted practice of clozapine augmentation. Results of studies examining the efficacy of antipsychotic augmentation of clozapine are equivocal or at best show modest effects [Taylor et al. 2012]. Polypharmacy compared with monotherapy is associated with, amongst other things, a higher rate of both metabolic adverse effects [Citrome et al. 2004; Correll et al. 2007] and extrapyramidal side effects [Carnahan et al. 2006] and reduced survival rates [Waddington et al. 1998].
The addition of a second antipsychotic to established monotherapy may, however, be justified in some cases. A meta-analysis of controlled trials of antipsychotic combinations versus monotherapy concluded that combinations may be beneficial in certain situations [Correll et al. 2009]. In addition, reports of the benefits of aripiprazole as an adjunct to clozapine treatment for the attenuation of metabolic adverse effects and raised prolactin are emerging in the literature [Fleischhacker et al. 2010; Fan et al. 2013]. Often in practice, however, despite the lack of robust evidence, combinations of antipsychotics, with or without clozapine, are prescribed in the absence of a clear alternative.
For our second intervention we compared antipsychotic prescribing across the trust. Prescribing practices and culture were found to differ between units, including those with a similar patient demographic. In addition, contrary to what we may have expected, prescribing high doses and combinations of antipsychotics was not limited to PICUs, nor did all PICUs prescribe high doses and combinations. These findings were presented at senior prescriber meetings. Through this process the perception that patient populations on certain wards were ‘more severely ill’ or ‘more difficult to manage’ was challenged. It was also possible to show that the use of high doses and combinations was by no means a certainty on PICUs.
There is no nationally or internationally agreed figure for what might be considered an acceptable rate of antipsychotic polypharmacy. The prevalence of treatment resistance is estimated at around 20–30% [Conley et al. 1997] and only around a third of patients will show an adequate response to clozapine alone [Kane et al. 1988]. Recent data suggest that patients may successfully be switched from polypharmacy to monotherapy without affecting symptom control or increasing the rates of hospitalization [Essock et al. 2011]. By using the formula X = 0.3 × 0.3 (assuming 30% of all our patients are treatment resistant and a further third have not responded adequately to clozapine treatment) we could expect around 10% of patients to be prescribed an antipsychotic to augment clozapine. However, the reports of the prevalence of treatment resistance in the general population vary, depending on the criteria used for identification; rates of up to 60% have been proposed [Essock et al. 1996]. Thus, accepting this higher estimate of nonresponse to standard treatment, for our third intervention we agreed a target for SLAM to reduce the rate of prescribing high-dose antipsychotics and polypharmacy to less than 20% (i.e. assuming 0.6 × 0.3 = 0.18 treatment resistance and nonresponse to clozapine).
Quality improvement programmes employ various strategies to influence change. Those which target barriers to change have been shown to be effective [Grimshaw et al. 2001; van Bokhoven et al. 2003]. For example, in hospitals medical hierarchy can influence local practice: junior staff may be more inclined to follow and less willing to challenge prescribing by senior colleagues. Understanding such behaviours and cultural beliefs of individuals can help to identify factors which delay or prevent change [Charani et al. 2011]. Interactive educational meetings, such as discussion and workshops, are more effective than didactic teaching sessions or electronic distribution of clinical materials. However, a multifaceted approach [Bero et al. 1998], such as participation in audit and provision of locally agreed guidelines, is better than any single strategy alone [Grimshaw et al. 2001; Grol et al. 2002]. Tailored approaches at various levels of an organization are generally required for any change [Grol and Grimshaw, 2003].
In 2009, we reported an increase in the proportion of women of child-bearing age prescribed valproate who were informed of its risks of teratogenicity. We also revealed that, although information provision improved, prescribing folate and contraception did not increase [Mace and Taylor, 2011]. Interventions aimed at changing prescribing practices in psychiatry have varied in their effectiveness to deliver such change. In 2001, Elliot and colleagues reported an improvement (reduction) in benzodiazepine prescribing in older people [Elliott et al. 2001]. In a UK-wide programme, however, strategies aimed at reducing the rates of prescribing high doses and combinations of antipsychotics were largely unsuccessful [Paton et al. 2008].
Our population is broadly representative of an ethnically diverse, inner-city hospital. For example, in our baseline survey, 44% of patients were white, 40% black and 16% from other ethnic origins; 74% had a primary diagnosis of a psychotic spectrum disorder and 67% were detained under the Mental Health Act. The results of our audit are interesting because they show a change in prescribing behaviour across a large organization. Rates of high-dose antipsychotic and combination prescribing are now lower in SLAM than those reported nationally [Prescribing Observatory for Mental Health, 2012]. The success of our programme may be attributed to the use of a combination of varied but precisely targeted strategies.
There is one noteworthy limitation to our study. Data in our audits were collected on a single day across the trust. As the audit is conducted annually, it is possible that prescribing practice improved temporarily in the weeks surrounding the audit.
We learned a number of important lessons during this programme. We found that a proposed change from established practice is often met with initial resistance. Any need for change and any proposed changes in practice should be discussed widely within the organization. In addition, endorsement from trust executive committees and the board, that is, a ‘top-down’ approach, proved to be particularly effective. Persistence with an improvement programme can continue to produce a sustained change, until a new culture is established: interventions introduced at each stage of our quality improvement programme are now part of routine clinical practice. Incremental improvements in practice should be noted; absolute targets of 100% achievement were seen as difficult to meet. ‘Benchmarking’ and ‘peer review’ were useful for identifying outlying practice and providing a healthy competition amongst prescribers.
In summary, our results show that it is possible to produce and sustain a change in prescribing culture across a large organization. Interventions should be specifically targeted to improve practice as highlighted at each stage of the audit. In order to achieve a change in culture, the target must continue to be viewed as a priority.
Footnotes
Funding
This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.
Conflict of interest statement
Mace – none. Professor Taylor has received consultancy fees, lecturing honoraria and research funding from AstraZeneca, Janssen-Cilag, Servier, Sanofi-Aventis, Lundbeck, Bristol-Myers Squibb, Novartis, Eli Lilly and Wyeth.