Benchmarking benzodiazepines and antipsychotics in the last 24 hours of life

The use of benzodiazepines and antipsychotic (phenothiazines and butyrophenones) medication in the final 24 hours of life is common within palliative medicine. These medications are used for a variety of indications, including nausea, delirium, anxiety, dyspnoea and seizures. However they are often grouped together as ‘sedative' medications, regardless of the intent with which they are being used, as a side effect for most of these can be a degree of sedation.Sedative drugs can be used as part of a targeted treatment to lower a patient's consciousness in a titrated, proportional way for relief from intolerable and refractory symptoms. This specific use is known as palliative sedative therapy (PST), terminal sedation or palliative sedation.Many ethical questions and controversy surround the use of PST but it has general acceptance within palliative medicine as an infrequently used end-of-life treatment option. Detailed guidelines are now available from the US and Europe,1,2 which includes the call for a regular review of practice for the purposes of improving quality of care (p921).1However BDZ/APS are frequently used at the end of life for many different and often much less clearly defined indications, including anxiolysis and terminal restlessness.3 For this latter indication, prescribing and administration practices may vary widely, and has both advocates and critics.4-7Where there is no universally agreed best practice, benchmarking is useful in enabling sometimes isolated health care providers ensure their practice is compatible with that of their colleagues8,9 and may help improve practice.10‘Normal' prescribing of BDZ/APS in NZ is not well documented although there are some guidelines.11,12 Given this possible variation, hospices in NZ volunteered to submit data to establish a benchmarking cycle of these medications at the end of life in NZ. In this context, the benchmark is not against a standard or a centre of excellence, but rather it establishes a range and distribution of practice within which a hospice can orientate and understand its own practice.This is the third cycle for benchmarking medication at the end of life, with data on opioid use previously published.13Method Study design—A cross-sectional design with retrospective chart review was used. All hospices with inpatient beds (n=18) identified through the Hospice New Zealand (HNZ) website were invited to take part with a written invitation. An open invitation was also circulated to all hospices through the HNZ e-mail list.. The written invitation included a document outlining the aims of the study and instructions on how to complete the electronic data collection template. Hospices were asked to undertake a retrospective review of notes and drug charts of consecutive patients who died under their care in February and March 2008. The expectation was that this would provide at least 20 patients for each unit, recognising some small hospices would not meet this goal, and larger hospices would exceed it. For admission to the study patients had to be under hospice care for at least the last 24 hours of their life and be 18 years or older. An accurate record of medications administered had to be available, which for many hospice teams confined the suitable patient pool to patients who had died in their Inpatient Unit (IPU). The study variables were age, gender, diagnosis (coded using the UK palliative care minimum data set14 with the addition of Melanoma as a specific diagnosis), use of the Liverpool Care Pathway (LCP) and place of death (Inpatient Unit (IPU), Aged Residential Care (ARC), public hospital or home). The template allowed for documentation of each medication used, the route of administration, and the total amount of medication administered in the 24 hours before death. For continuous subcutaneous medication or long-acting medication which was either started or changed during the last 24 hours, pro-rata calculation was used to ascertain the actual dosage delivered. Other medications given for symptom management and any comments thought to be relevant were also recorded. The templates were collated, and the initial analysis shared with the contributing hospices for comment and correction. The final analysis was returned to each hospice containing their own analysed data, enabling them to compare medications and dosages used with the rest of NZ. Analysis—Data was anonymised by allocating each hospice a number, and entered into PASW (Previously SPSS) Version 18.0 for statistical analyses. This included descriptive analysis, then analysis of variance looking for differences in practice between hospices. A log transformation of the dosage data was used to more approximate a normal distribution, and to exclude ‘nil' dosages from the calculation of median dosages. A Haloperidol Equivalent Daily Dosage (HEDD) as described by Hui15was used to combine haloperidol and levomepromazine (NozinanTM) as a single dosage figure. (HEDD=total haloperidol dosage (mg) by any route + oral levomepromazine dosage (mg) × 8/300 + parenteral levomepromazine dosage (mg) × 8/100.) To facilitate an easy visual comparison between hospices, a drug ‘footprint' was devised for both benzodiazepines and antipsychotics which records the percentage of patients in each hospice receiving a particular medication or no medication. Only the occurrence of use is displayed, regardless of the frequency or dosage of each medication administered to a patient. Ethics—A certificate of approval was obtained from the New Zealand Health and Disability Ethics Committee (Central Regional Ethics Committee). Further ethical approval was not required for this retrospective study. Results Fourteen hospices across rural and urban NZ contributed data on 351 patients. There were 181 males and 170 females, with a mean age of 70 (range 21-96). 64 patients died at home, 6 in a public hospital, 251 within a hospice inpatient unit and 30 in an ARC setting. (Figure 1). Figure 1. Percentages of patient data for each place of death over 14 hospices 302 patients (86%) died of cancer-related diseases and 49 (14%) died of non-malignant disease. The most common cancers were gastrointestinal (24%), lung (17%), prostate and breast (both 7%) and melanoma (6%), while notable non-malignant diseases included chronic respiratory disease (4%) and heart failure (3%). Benzodiazepines included midazolam, clonazepam, and flunitrazepam, with single uses of diazepam and lorazepam. Antipsychotics were commonly haloperidol and levomepromazine. Quetiapine and risperidone were used infrequently. Phenobarbitone was used once. Separate footprints for benzodiazepines and for antipsychotics have been generated (Figures 2 and 3). Seventy-five percent of patients overall received a benzodiazepine. Midazolam was the most commonly used, although flunitrazepam was preferred at one hospice. The percentage of patients in each hospice receiving either midazolam or flunitrazepam in the last 24 hours of their life varied from 40%-87%. If any benzodiazepine was included, percentages ranged from 45%-93% (Figure 2). Up to 39% of patients at each hospice received more than one benzodiazepine, with a median of 5%. The combination of benzodiazepines was always midazolam and clonazepam apart from two instances of midazolam and flunitrazepam. Figure 2. Benzodiazepine footprint Figure 3. Antipsychotic footprint Overall 62% of patients received an antipsychotic, with levomepromazine used more frequently than haloperidol (147 vs 94 patients). There was a range in the incidence of the use of any antipsychotic from 30% up to 97% between hospices (Figure 3). Up to 35% of patients at a hospice received the combination of both haloperidol and levomepromazine with a median of 2.5%. About three-quarters of patients receiving levomepromazine were administered 12.5 mg or less in 24 hours (74%), 13% received over 25 mg. Dosage of medications—Table 1 shows the range, median and mean for dosages of the five principal medications. Table 1. Dosage descriptors Medication Range (mg) Median (mg) Geometric mean (mg) Midazolam Flunitrazepam Clonazepam Haloperidol Levomepromazine 1-118 0.5-96 0.125-5 0.1-10 0.1-275 15 12 1 2.5 12.5 20.7 20.5 1.3 2.73 21 When comparing dosages across patients who died in the different hospice IPUs, there is a significant variation between the 14 institutions in midazolam and HEDD dosages (Table 2). Table 2. ANOVA using hospice as dependant factor Variables df F value P value Midazolam All deaths Deaths in IPU 208 156 1.971 1.880 0.025 0.037 HEDD All deaths Deaths in IPU 219 162 4.203 1.916 <0.001 0.032 HEDD=haloperidol equivalent daily dosage; IPU=Inpatient Unit; df=degrees of freedom. Table 3. (Bivariate) comparisons of midazolam dosage Midazolam Geometric mean Midaz (mg) N df F value P value IPU death 14.8 157 208 7.950 0.005 Death not in IPU 9.8 52 Male 13.1 106 208 0.050 0.823 Female 13.6 103 Malignant (IPU) 15.2 135 156 0.847 0.359 Non-malignant (IPU) 12.5 22 Died on LCP* 15.3 46 156 10.824 0.001 Died off LCP* 9.5 24 Died in IPU using LCP