Abstract
The use of ‘as needed’ or pro re nata (PRN) psychotropic medications has been common in psychiatric inpatient care for many years [1–4], but traditional practices need to adjust both to changes in the spectrum of illness, and the advent of better pharmacological and non-pharmacological treatments. Traditionally, typical antipsychotics such as chlorpromazine or haloperidol were used to sedate acutely disturbed psychotic patients. However, these drugs cause significant extrapyramidal symptoms (EPS) such as parkinsonism, acute dystonia or akathisia in 50–75% of patients [5, 6]. More modern approaches to the scheduled pharmacotherapy of psychoses entail the use of atypical antipsychotics or lower doses of typical antipsychotics [7].
Safety and efficacy of PRN medication
Although PRN medications are essential for the management of agitation, they may also be the direct cause of adverse reactions such as EPS or excess sedation [8]. The use of antipsychotics for the acute control of agitation relies predominantly on their sedative side-effects rather than on any immediate effect on the psychotic disorder. This direct effect takes several days or even weeks and requires ongoing scheduled administration of low to moderate doses of antipsychotic drugs [7].
There is a paucity of outcome studies on the use of oral PRN psychotropic medications in psychiatric patients. The few studies that concluded that this practice was effective relied on assessments made by those administering the medication and were hampered by inadequate levels of documentation [2, 9, 10]. The most extensive literature exists for the procedure called ‘rapid tranquillization’, which involves the administration of parenteral medication to severely agitated individuals. Randomized controlled trials of rapid tranquillization have consistently demonstrated that antipsychotics and benzodiazepines are equally effective for the acute sedation of agitated patients, and that acute psychotic symptoms respond equally well to either agent. However, antipsychotics showed greater toxicity than benzodiazepines [11–14]. Thus for agitated psychotic patients, benzodiazepines are recommended as first line adjuvant treatments [13, 15]. This position recently received support from the RANZCP Committee for Psychotropic Drugs and Other Physical Treatments [16]. In practice, however, approaches to the use of PRN psychotropic medication vary widely between institutions, prescribers and those who administer the medications.
Anticholinergic medications are often given PRN to treat or prevent EPS. However, their prophylactic use is controversial and not all EPS respond to anticholinergics. While parkinsonism and acute dystonias respond well to anticholinergic medications for akathisia, propanolol or benzodiazepines or reduction of antipsychotic dosage are more likely to be effective [17, 18]. Tardive dyskinesia does not respond to anticholinergics and may even worsen [19].
The decision-making process
The decision-making processes involved in the use of PRN medication are complex. Doctors (particularly junior doctors) often prescribe at the behest of nursing staff. Frequently, prescriptions for several medications are given concurrently, thereby allowing nurses to choose which medication to administer. Many individual prescriptions lack key specifications such as an indication for administration or a maximum daily dose, thereby forcing nurses to make these decisions. Furthermore, it is common practice to offer a dose range for each administration, (e.g. chlorpromazine 50–100 mg prn) [8, 9]. Thus, nurses not only initiate most episodes of PRN administration but also make many decisions on the choice of agent/s, the doses and the reasons for administration. The extent to which doctors can abrogate their responsibility for these decisions is open to debate.
Patients, doctors and nurses often perceive the use and effects of psychotropic medications differently [20]. We therefore conducted a survey to register the knowledge and beliefs about PRN medication of the medical and nursing staff of two major metropolitan mental health inpatient services. Since surveys of clinical behaviour may show bias towards idealized responses, we also undertook a large retrospective chart audit of PRN psychotropic medication for psychotic disorders that we report separately [21].
Method
A written questionnaire comprising of 15 questions on knowledge and beliefs about PRN medications was administered to staff of the inpatient psychiatry units of two major metropolitan hospitals. The questionnaire took about 10 minutes to complete. It contained three questions where respondents were asked to select indications for administration of PRN medication, and two questions where respondents were asked to select from a range of medications, which they believed were effective for agitation and psychotic symptoms. Four open-ended questions on preferred and specifically avoided medications for these symptoms had follow-up questions asking respondents to give a rationale for their choice. Two open-ended questions were asked on methods used to assess response to PRN medication. Two questions asked respondents to describe alternatives to PRN medication for these symptoms and a further two asked how often they used these alternatives. The survey was conducted over a four-month period from March–June 1999, one month following an audit of PRN medication practice in both units [21].
No similar instruments were available for cross-validation, but feedback from psychiatrists, senior nurses and pharmacists was used to enhance the survey's face validity. A small pilot survey led to the modification of one question, but overall showed the questionnaire to be easily understood and completed. Staff were recruited directly using opportunistic methods such as visiting wards, journal clubs, hand-over meetings and other staff meetings. Resource limitations and a likely higher response rate were factors in choosing this potentially biased method, over more indirect approaches. The survey was completed in the presence of the principal investigator (JG). Conferring was discouraged although points of clarification could be sought from the investigator.
Approval for the study was obtained from the Clinical Research Ethics Committees of the Royal Brisbane and Princess Alexandra Hospitals. The data were entered into a database and analysed using Access © and SPSS © version 9.0 software packages. Continuous variables were examined using either the student's t-test or ANOVA, with post hoc comparisons using Tukey's Honestly Significant Differences test. Categorical data were examined using χ2 and Odds Ratio (OR) tests.
Results
Participation rates
Ninety-eight per cent (127/129) of the staff members who were approached agreed to participate. Of the total staff of the two units, 61% (80/132) of nurses (6 clinical nurse consultants, 17 clinical nurses and 57 registered nurses), and 89% (47/53) of doctors (8 consultants, 31 registrars and 8 residents) completed the survey.
Beliefs about indications for PRN medication
Staff were asked to select from a range of symptoms which they believed were appropriate indications for the use of PRN psychotropic medications. More than one symptom could be selected. The numbers of doctors and nurses who nominated specified symptoms as indications for PRN antipsychotic, benzodiazepine, and anticholinergic administration are shown in Tables 1 and 2.
Beliefs about appropriate indications for administering pro re nata (PRN) antipsychotics or PRN benzodiazepines by nurses (n = 80) and doctors (n = 47)
Beliefs about appropriate indications for pro re nata anticholinergics by nurses (n = 80) and doctors (n = 47)
Antipsychotics
Nurses nominated more indications (mean = 3.49) than did doctors (mean = 2.72) (t = −2.37, df = 125, p < 0.05; Mean Difference = −0.76, 95% CI = −0.13 to −1.40). A high proportion of both staff groups felt that agitation was an appropriate indication for PRN antipsychotic use (doctors 70%, nurses 74%). A smaller proportion of doctors (45%) and a very high proportion of nurses (93%) felt that hallucinations and/or delusions were appropriate indications for PRN antipsychotic use. Formal thought disorder was frequently selected by nurses (50%) but not doctors (17%). Less conventional indications for PRN antipsychotics that were often selected included anxiety, sleep disturbance, dysphoria, suicidal ideation, and conflict with copatients.
χ2 analyses showed that nurses were significantly more likely than doctors to select hallucinations and/or delusions (χ2 = 35.91; OR = 15.27, 95% CI = 5.59–44.31, p < 0.001), formal thought disorder (χ2 = 13.67; OR = 4.88, 95% CI = 2.05–12.26, p < 0.001), and suicidal ideation (χ2 = 10.70; OR = 5.04, 95% CI = 1.85–15.61, p < 0.01) as appropriate indications for PRN antipsychotic administration. There was no significant difference in the rates at which the two staff groups selected other indications for PRN antipsychotics.
Benzodiazepines
Doctors nominated significantly more indications (mean = 3.77) for PRN benzodiazepine administration than nurses (mean = 3.19) (t = 2.15, df = 125, p < 0.05; Mean Difference = 0.58, 95% CI = 0.04– 1.11). A high proportion of both staff groups believed that anxiety (doctors 89%, nurses 95%), agitation (doctors 96%, nurses 78%) and sleep disturbance (doctors 91%, nurses 58%) were appropriate indications for PRN benzodiazepines. Dysphoria, suicidal ideation, and conflict with copatients received intermediate levels of support as indications for PRN benzodiazepines. Few staff felt hallucinations and/or delusions were appropriate indications for PRN benzodiazepines.
χ2 analyses demonstrated that doctors were significantly more likely than nurses to select agitation (χ2 = 7.43; OR = 6.53, 95% CI = 1.43– 60.21, p < 0.01, Fisher test), and sleep disturbance (χ2 = 16.31; OR = 7.95, 95% CI = 2.49–32.91, p < 0.001, Fisher test) as appropriate indications for PRN benzodiazepine use. No differences emerged in the rates at which the two groups nominated other indications.
Anticholinergics
Doctors and nurses did not differ in the number of indications nominated for anticholinergics. A high proportion believed that acute dystonia (doctors 96%, nurses 85%), muscle stiffness (doctors 64%, nurses 70%) and tremor (doctors 38%, nurses 33%) were appropriate indications for giving PRN anticholinergics. Bradykinesia, defined as slowed movements was more likely to be selected by doctors than by nurses (χ2 = 13.24; OR = 4.17, 95% CI = 1.87–9.29, p < 0.001). In contrast, akathisia was more likely to be selected by nurses than by doctors as an indication for PRN anticholinergic administration (χ2 = 4.92; OR = 2.38, 95% CI = 1.08–5.19, p < 0.05).
Inappropriate indications for anticholinergic agents, including tardive dyskinesia (doctors 19%, nurses 38%) and cholinergic blockade side-effects such as dry mouth or blurred vision (doctors 4%, nurses 20%), were selected by a number of respondents. Nurses selected these symptoms at a significantly greater rate than doctors; tardive dyskinesia (χ2 = 4.69; OR = 2.53, 95% CI = 1.08–6.12, p < 0.05) and blurred vision (χ2 = 6.03; OR = 5.63, 95% CI = 1.21–52.29, p < 0.05, Fisher's exact test).
Perceived effectiveness of various PRN medications
Respondents were asked to select which of a given range of PRN medications would be effective for agitation. Multiple drug choices were allowed. A similar question asked which PRN medication respondents believed would be effective in reducing psychotic symptoms. Responses are shown in Table 3.
Belief in the effectiveness of specific pro re nata medications by nurses (n = 80) and doctors (n = 47)
There was broad support for the effectiveness of PRN conventional antipsychotics for either psychotic symptoms or agitation (< 80% of all staff). Of the antipsychotic drugs, chlorpromazine was most likely to be selected as effective. Pro re nata benzodiazepines were regarded as effective for agitation (doctors 94%, nurses 78%) but less so for psychotic symptoms (doctors 32%, nurses 58%). Clonazepam 1 and diazepam were the benzodiazepines most likely to be selected as effective.
Preferred choice of PRN medication
Open-ended questions asked staff to nominate their preferred choice of PRN medication/s for managing agitation and to explain the reason for their response. Nurses were more likely to nominate an antipsychotic than doctors (60% vs 30%, χ2 = 11.08; OR = 3.56, 95% CI = 1.65–7.71, p < 0.01). Chlorpromazine and haloperidol were the most frequently cited. In contrast, doctors were more likely than nurses to prefer a benzodiazepine (55% vs 30%, χ2 = 11.78; OR = 3.71, 95% CI = 1.71–8.04, p < 0.01). Diazepam and clonazepam were the most frequently cited. When asked which PRN drug they would avoid for agitation and why, oxazepam was cited most often usually because of a perceived ‘lack of effectiveness’. Next was haloperidol, with ‘sideeffects’ being the most frequently cited rationale for avoiding its use.
Similar open-ended questions asked staff about their preferred PRN medication for managing psychotic symptoms. Nurses were more likely to nominate an antipsychotic than doctors (87% vs 64%, χ2 = 8.66; OR = 3.55, 95% CI = 1.47–8.65, p < 0.01). Haloperidol, the most frequently cited choice, was nominated by nurses and doctors at a similar rate (51% vs 45%). Nurses cited chlorpromazine as often as haloperidol, but fewer doctors did (51% vs 11%, χ2 = 20.05; OR = 8.40, 95% CI = 3.11–25.75, p < 0.001). A subgroup (15% of doctors and 3% of nurses) felt that no PRN medication was their preferred choice in the face of psychotic symptoms on the grounds that ‘PRN medication is generally not effective’. When asked which PRN drugs they avoid in treating psychotic symptoms, oxazepam, diazepam and clonazepam were cited at similar rates. The principal reasons given were that these drugs ‘lack antipsychotic effect’ and ‘risk of dependence’.
Assessment of effectiveness of PRN medication
Respondents were asked to describe how they assess whether a PRN medication has had the desired effect. More than half of doctors (54%) and nurses (61%) nominated a combination of subjective (e.g. inquire about the patient's internal state) and objective means (e.g. observe the patient's behaviour) of assessment. One-fifth of doctors and nurses nominated only one type of assessment (subjective or objective). No doctors and only 4% of nurses specifically included an assessment of side-effects in their responses.
Alternatives to PRN medication
Staff were asked to nominate up to four alternatives to the use of PRN medication. Staff cited significantly more alternatives for the treatment of agitation (mean = 3.07, SEM ± 0.10) than for the treatment of psychotic symptoms (mean = 2.24, SEM ± 0.13) (paired t = 6.72, df = 126, p < 0.001; Mean Difference = 0.83, 95% CI = 0.58– 1.07). Nurses nominated more alternatives for treatment of agitation than did doctors (nurses: mean = 3.24, SEM ± 0.12; doctors: mean = 2.79, SEM ± 0.19) (t = −2.14, df = 125, p < 0.05; Mean Difference = −0.45, 95% CI = −0.03 to −0.87), but groups did not differ in alternatives nominated for treatment of psychosis (nurses: mean = 2.37, SEM ± 0.16; doctors: mean = 2.34, SEM ± 0.19).
One-way analyses of variance showed that, among nursing staff, the effect of seniority on the number of alternatives nominated was significant, F (2, 77) = 4.461, p < 0.05. Post hoc testing with Tukey's HSD test showed that registered nurses nominated significantly fewer alternatives than clinical nurse consultants (Mean Difference = −2.10, 95% CI = −0.03 to −4.17, p < 0.05). Among doctors, an effect of seniority on the number of alternatives to PRN medication nominated was also evident, F (2, 44) = 4.367, p < 0.05, with post hoc Tukey testing indicating that residents nominated fewer alternatives than consultants (Mean Difference = −3.13, 95% CI = −0.17 to −6.08, p < 0.05) and registrars (Mean Difference = −2.67, 95% CI = −0.32 to −5.02, p < 0.05).
The most commonly cited alternatives to the use of PRN medication were counselling, distraction, time-out, reality testing, relaxation and cognitive behaviour therapy. Nurses reported a significantly higher frequency of use of these techniques than did doctors. Over the week prior to surveying, nurses reported using alternatives to PRN medication for agitation on an average of 21.7 occasions (SEM ± 3.6) and doctors on an average of 3.2 occasions (SEM ± 0.6) (t = −3.88, df = 125, p < 0.001; Mean Difference = −18.50, 95% CI = −9.03 to −27.96). For psychotic symptoms, non-pharmacological interventions were used by nurses on 17.7 occasions (SEM ± 3.5) and by doctors on 3.4 occasions (SEM ± 0.9) in the previous week (t = −3.12, df = 125, p < 0.01; Mean Difference = −14.30, 95% CI = −5.23 to −23.37).
Discussion
Opportunistic sampling was used because of limited resources and to achieve a higher response rate. Even though a high proportion of staff were surveyed, this may have introduced bias. Visiting medical officers and night staff were two groups less likely to be surveyed. The presence of the principal investigator may have lead to a bias in favour of idealized responses, a significant issue when surveying clinical beliefs. Finally, the results were obtained from two inpatient units in the same city and thus may have limited generalizability to other settings.
The disparities between the views of nurses and doctors about appropriate indications for administering PRN medication are of particular concern when doctors frequently fail to specify indications for administration on PRN prescriptions [3, 10]. This disparity was most marked in the area of psychotic symptoms where nurses were far more likely than doctors to consider hallucinations/ delusions or even formal thought disorder as an indication to administer PRN antipsychotics. Hence, patients will receive medication for reasons that were not intended by the prescribing doctor.
A significant number of respondents nominated indications that are not in keeping with the approved indications or the known properties for that class of drug. Some 34% of doctors and 29% of nurses stated they would give PRN antipsychotics for either sleep disturbance or anxiety. The use of PRN antipsychotics to treat formal thought disorder (nominated by 50% of nurses and 17% of doctors) is puzzling as most patients are not overtly distressed by formal thought disorder. Furthermore, even if the formal thought disorder was distressing to the patient, it is doubtful that PRN antipsychotics are acutely effective.
Beliefs about anticholinergic indications were also of concern. Some doctors and many nurses surveyed would use anticholinergics to treat tardive dyskinesia even though evidence suggests they offer no benefit and may even worsen symptoms [19]. Furthermore, 23% of nurses and 6% of doctors stated they would give PRN anticholinergics to treat dry mouth or blurred vision, thereby exacerbating these symptoms.
Recent research shows that the short-term response of psychotic symptoms to PRN benzodiazepines is equivalent to that observed with PRN antipsychotics and that benzodiazepines are safer, better tolerated and avoid problems related to antipsychotic polypharmacy [11–15]. Yet only 32% of doctors believed that PRN benzodiazepines are effective against psychotic symptoms, and there was a clear reluctance on the part of both doctors and nurses in the present study to use benzodiazepines for acute psychotic symptoms. The results from questions on indications and perceived effectiveness of PRN medications suggest that many staff are unaware of (or doubt) the acute effectiveness of PRN benzodiazepines against psychotic symptoms.
While 94% of doctors and 78% of nurses believe PRN benzodiazepines are effective for agitation, approximately 60% of nurses and 30% of doctors prefer to use PRN antipsychotics for this indication. This is despite the evidence that in acute settings benzodiazepines are relatively safe and effective agents with limited potential for abuse in contradistinction to chronic administration where they are less effective and more likely to result in dependence syndromes [22]. This situation is akin to the use of opiates for analgesia. Opiates have a low potential for dependence in acute settings and few people argue against the use of opiates in acute pain. When patients are given control of their own analgesia in acute pain settings they use less rather than more medication [23, 24], and it is an interesting question whether patients with psychoses requiring PRN treatment for agitation arising from acute ‘psychic pain’ would behave differently if given a measure of control over their medication. Like opiates, care would be needed to ensure that they were weaned as acute symptoms improve.
PRN medication outcomes are poorly documented in clinical practice [9, 10, 21], even though most staff were able to nominate both objective and subjective methods for assessing the outcome of PRN treatment. Thus, failure to record outcomes is probably a consequence of a systemic failure rather than a lack of staff knowledge of how to assess outcome. One suggestion to overcome this recording failure is to insert a ‘PRN medication stamp’ or ‘sticker’ in the clinical notes for each occasion of administration that includes prompts for recording the indication, dose administered, outcome and side-effects. This would also allow easy tracking of patient PRN requirements and inform scheduled prescribing.
Both nursing and medical staff, particularly more senior staff, were able to nominate many non-pharmacological techniques for managing both psychotic symptoms and agitation. The literature in this area is abundant and many of the techniques have been found to be successful [25]. The lower levels of knowledge among junior staff suggests a need for further attention to this area during basic training of both doctors and nurses, and for in-service training to further develop these skills.
In the present study, one unit relied more heavily on casual staff, and had an older ward design without high dependency areas than the other. The nurses at this unit frequently commented that they were hampered in their ability to use non-pharmacological treatments by the ward environment and staff skill levels. This illustrates that a significant determinant of practice change is management of the work environment in addition to addressing staff attitudes or knowledge [26].
Conclusions
While nurses and doctors agree on many aspects of the use of PRN medications in acute psychoses, there are key areas of difference. Disparities exist between doctors and nurses primarily on the appropriate indications for using PRN medications. Thus, it is important for doctors to specify indications when writing PRN prescriptions.
Despite evidence for their safety and effectiveness there is widespread reluctance to use benzodiazepines as PRN medication. Barriers to the acute use of these agents, which are not in keeping with recent research findings, include doubts about their efficacy and concerns about inducing dependence. Monitoring outcomes of PRN medications, including side-effects may assist in challenging the beliefs which underlie clinical practice.
These results suggest that many nurses and doctors in psychiatric settings could benefit from further education about psychotropic medication. This is in keeping with aged care research suggesting more than three quarters of nurses felt their basic nursing education regarding the effects and side-effects of psychotropic medication was inadequate [27, 28]. Junior staff would also benefit from more training in non-pharmacological management skills.
Footnotes
Acknowledgements
We thank the Royal Australian and New Zealand College of Psychiatrists, the Princess Alexandra Hospital Research & Development Foundation, the Lions Medical Research Foundation, the Australian National Health & Medical Research Council and the Danish Research Academy. We thank the Royal Brisbane Hospital, Princess Alexandra Hospital Divisions of Mental Health and participants in the study.