Factors influencing prescription and administration of analgesic medication: A longitudinal study of people with dementia living in care homes

To (1) describe the prescription and administration of regular and ‘as required’ (pro re nata [PRN]) analgesics in English care homes, (2) investigate individual and care home factors associated with analgesic use.


| INTRODUCTION
As the United Kingdom (UK) population ages, increasing numbers of older people will spend their last months of life in a care home. 1 These provide accommodation, meals and 24-hour staffing. There are two types in the UK: residential homes, which provide assistance with personal care, and nursing homes, which are similar to residential homes but also employ registered nurses round-the-clock. It is estimated that around 80% of care home residents have some form of dementia or severe memory problem. 2,3 Around half of care home residents are in pain and there are concerns that this may be undertreated, particularly in residents with cognitive impairment. 4,5 Normally, medication is prescribed by general practitioners and administered by care home staff.
Paracetamol is indicated as first-line therapy for mild to moderate pain in older adults. 6,7 Nonsteroidal anti-inflammatory drugs (NSAIDs) are effective in relieving musculoskeletal pain, but are not advised in older adults due to risks including gastrointestinal bleeding and effects on the renal system. 8 Opioids are the strongest class of analgesic, and are sometimes used long-term to treat severe pain.
There has been an increase in use of analgesic medication; especially opioids and paracetamol in care home residents over recent years. 9,10 However, this does not necessarily mean that pain management is improving: pain prevalence in this population may be rising due to increasing levels of multimorbidity. 11 Contemporaneously, there has also been increasing concern about opioid side effects (including dizziness, constipation, nausea, sedation and falls) 7 and whether opioids are being overprescribed in this population.
A pan-European study reported that 48.4% of nursing home residents experienced pain. 12 Data from England found pain prevalence to be 54.5%, with 8.1% of all residents in constant pain, 57.4% experiencing intermittent pain, 6.2% experiencing a single episode of pain and 10.9% who had experienced breakthrough pain. Of those in pain, 53.8% received regular analgesic medication, 16.3% were prescribed 'as required' (pro re nata [PRN]) analgesics only, and 4.3% received regular and PRN. 13 The WHELD study, conducted in England, found that 41.9% of residents in pain were not prescribed regular analgesics. 14 Research exploring analgesic use and individual differences presents a mixed picture. Historically, males, younger residents and those with a dementia diagnosis have received fewer analgesic prescriptions. 9,10 PRN analgesics that are prescribed but not administered, which is often the case, 15 may indicate undertreated pain. In England, it is not possible to accurately measure care home residents' receipt of PRN analgesics as there are no routine datasets or analgesic administration data accessible to inform this. 16 A negative association has been found between PRN administration and dementia severity suggesting that as people become less able to ask for analgesia they are more likely to have untreated pain. 17,18 Larger care homes prescribe more medication as PRN, 19 and communication issues between care staff have been implicated in quality of pain management. 20 A large European study found that factors that may contribute to poor communication, including heavy workload, high staff turnover and team instability, were associated with inadequate pain management. 13 The Managing Agitation and Raising QUality of LifE in dementia (MARQUE) study is the largest prospective study of English care home residents to date. This study utilised MARQUE data to describe analgesic medication use in a large and representative cohort of English care home residents. Our objectives were to: (1) describe the prescription and administration of analgesic drugs and drug classes, and (2) understand which care home or resident factors were associated with analgesic prescription or PRN administration. Residents were eligible if they had a recorded dementia diagnosis, or if they screened positive on the Noticeable Problems Checklist (NPC), 22 a scale validated against clinical diagnosis of dementia. All residents with dementia were invited to take part. We were guided by the UK Mental Capacity Act (2005) 23 when obtaining written consent. If care home staff deemed it likely that the resident would have capacity to consent to research then we approached the resident. If not, then the care home approached the next of kin to request permission for a researcher to make contact, and if they gave permission we sought consultee opinion. In cases where there was no next of kin, a care home staff member who knew the resident well, acted as a professional consultee.

| Measurement instruments
We recorded care home baseline characteristics (care provision; ownership; registration/specialism; number of beds; Care Quality Commission (CQC) rating, and resident demographics at baseline.
Care home staff members gave proxydata regarding dementia severity (Clinical Dementia Rating 24 [CDR]) at baseline, 4-month and 12-month study visits. The CDR is a valid and reliable measure of dementia severity generating a global impairment score (from 'none' to 'severe').

Medication prescriptions were obtained from Medication
Administration Records. If the prescription was PRN then the number of times it was administered during the preceding 2 weeks was noted. We analysed data regarding oral analgesics, categorised by British National Formulary 25 (BNF) classification: simple nonopioids, opioids and NSAIDs.

| Statistical analyses
Data were analysed using STATA 14. 26 For analysis, residents with questionable or very mild dementia were subsumed into the mild dementia group.
Compound drugs were divided into their constituent parts. For each drug and drug class, the number and percentage of residents who were prescribed each drug at baseline and 12 months was reported. For PRN prescriptions (where 14 days' worth of data were available), the percentage of doses administered, of the total potential amount prescribed, was calculated. Multilevel logistic regression models were used to explore resident factors (demographics; dementia severity) and care home factors (demographics; quality rating [CQC]) on: overall analgesic prescription; regular analgesic prescription; PRN analgesic prescription; PRN administration if prescribed a PRN analgesic (as binary [yes/no] variables). The three-level regression models accounted for clustering by care homes and repeated measures (baseline; 4 months; 12 months) within participants. Oral and nonoral nonmorphine opioids were converted to an equianalgesic dose of oral morphine using BNF conversion rates. 27 Prior to this, a sensitivity analysis was run to explore any baseline differences between residents who had died and those who were still alive (χ 2 for binary variables; t-test for continuous) with missingness following death at subsequent visits as an outcome. Mann-Whitney U tests were run for nonparametric data. If any variables predicted missingness they were included as independent variables in the models, as were factors with a clinical reason to do so.
A multilevel linear regression model explored the effects of care home factors (demographics; CQC ratings) on prescriptions and administration at baseline. Prior to this, univariate analyses to identify potential confounders were run, which were included in the final model. If no factors had a relationship with prescribing or administration, then heterogeneity between care homes was quantified by calculating I 2 (I 2 > 75% is considerable heterogeneity) and forest plots were generated.

| Demographics
Out of 114 homes approached, 86 participated. Two homes withdrew at the 4-month study visit, and a further two withdrew by the 12month study visit ( Figure S1). Compared to the national average, a higher proportion of dementia-registered and dementia-specialist homes, and homes rated as 'Outstanding' and 'Good' were recruited. Table S1 describes the consented care home characteristics.

| PRN administration
Across all three study visits, 41.9% of residents with a PRN prescription did not receive any PRN analgesics during the 2-week periods prior to assessment. Figure S2 describes the prescription and administration of PRN paracetamol and opioids at baseline. At baseline, 41.1% of residents did not receive any analgesic medication (none prescribed, or PRN not administered), and at 12 months this was 54.9%. There were 19 residents who received all of the potential analgesic doses across the study visits, and 344 residents who did not receive any. Table 3 describes resident factors associated with the prescription and administration of analgesics. Females were prescribed significantly more regular analgesics, but there were no sex differences in PRN administration. Tables S2 and S3 describe resident factors and their association with nonopioid and opioid prescription and administration. Residents with severe dementia were prescribed more regular opioids; they were no less likely to be prescribed PRN but were administered significantly fewer PRN analgesics than residents with mild dementia.

| Factors associated with analgesic use
The only care home factor significantly associated with prescription of analgesics (at baseline) was if the care home was rated as 'requires improvement' (Table S4). Residents in these care homes were less likely (odds ratio = 0.26; 95% confidence interval = 0.09, 0.92) to receive a prescription for an analgesic compared to residents living in a care home rated as 'Outstanding'. No care home factors were significant when exploring analgesic administration. Post hoc tests of heterogeneity were run to further explore differences. There was no heterogeneity in the analgesic prescribing amongst the care homes (I 2 = 0.0%), but there was moderate heterogeneity in the PRN administration (I 2 = 56.7%). These differences were not associated with the variables that were measured.

| DISCUSSION
This is the most comprehensive study in the UK to examine regular and PRN analgesic prescribing and administration in a care home population. Less than a third of residents in this cohort were prescribed regular pain relief; for 4 in 10 residents, pain management was reliant on PRN medication that was given less than half the time.
An increase in the use of strong opioids was observed over the three study visits. Most strong opioids prescribed in this cohort were T A B L E 1 Resident characteristics at baseline (n = 1425)  Abbreviations: PRN, pro re nata; NSAIDs, nonsteroidal anti-inflammatory drugs.

Characteristic Analgesic prescription Regular prescription PRN prescription PRN administration
Odds ratio (OR) or coefficient (coef.) (95% confidence intervals) Female sex (OR) a 1.27 (1.06, 1.53  raises the question whether opioids are being used, instead of antipsychotics, to manage behavioural symptoms. 28 Patches may be prescribed due to ease of administration 29 or because residents have swallowing difficulties. It is important to note that the difference between strong and weak opioids is somewhat arbitrary: a low dose of buprenorphine is less potent than a high dose of codeine, for example. Typically, residents were prescribed low doses of buprenorphine (5 mcg/h).
Regarding care home differences, analgesic prescriptions did not vary between care homes, but care homes varied considerably in how much they administered PRN analgesics, although it was not clear why. Previous literature has found that factors such as heavy workload and shift work which affect communication between care staff are associated with poor pain management. 12,20 Approximately 50% of care home residents have pain, and studies that collect data on pain and analgesia in care homes consistently find evidence that this is undertreated. 4,5,13 In our study, drug receipt varied between groups. Females were more likely to be prescribed regular analgesics and younger residents were administered more PRN analgesics. Clinical studies suggest females may experience more pain, and report pain more frequently. 30,31 Residents with severe dementia were more likely to be prescribed regular analgesics but less likely to be administered PRN analgesia, despite being no less likely to be prescribed PRN analgesics. Clinicians prescribe PRN analgesia with the expectation that it may be needed; so the finding that being prescribed PRN analgesia that is not administered is associated with having more severe dementia and being older, which are characteristics that might be expected to be associated with greater risk of pain, could lead to undertreated pain. Care home staff may find it more difficult to assess pain in residents with cognitive impairment, especially if residents have difficulties communicating, 32 and undertreatment of pain has been consistently found in this population. 10,14 Our findings suggest that while overall analgesic prescribing is similar to other countries, clinicians in England prescribe more PRN analgesics than in other countries where regular pain relief is more commonly used. 33 It is not possible to speculate on the appropriateness of the PRN prescriptions from our study, but assuming that care home populations in other countries are broadly similar, it appears English care home residents are receiving less pain relief than their international counterparts. UK opioid prescriptions in older people residing in the community in their own homes and care homes were similar in 2010 (22.4% vs. 20.1%) 34 and in our cohort, prevalence was only slightly higher (22.8%-23.6%), which reflects the upward trend seen elsewhere. 35 Therefore, our finding of less prescription of analgesics in English care homes, compared to international levels, may reflect the wider national healthcare system.

| Strengths and limitations
Earlier studies have commented on the need for PRN administration data 15,16,18 and this study has presented robust data from a large sample that is representative of a heterogeneous group. The results aid interpretation of the limited existing medication data and highlight the importance of collecting data on frequency of medication administration. The inclusion criteria were pragmatic, and the use of the NPC meant that residents with dementia that did not have a formal diagnosis were able to participate.
The medication charts did not routinely include the indications for prescriptions, so adjuvant drugs such as amitriptyline or pregabalin prescribed for specific types of pain relief were not included in the analyses. Nor did it collect data about 'home remedies': medications such as paracetamol may not require prescription but are administered by care home staff.
Without pain data it was impossible to assess the appropriateness of prescriptions, type of analgesic, or administration. However, pain can be difficult to assess in this population, and fluctuates over time, so pain assessments may not be accurate. 36 Given that care homes and their residents are so heterogeneous, future studies need to balance a large sample size with pain assessments and detailed medication data. Where pain assessment is not feasible, other options include records of potentially painful comorbidities or a medication-based comorbidity index. [37][38][39][40] To assess whether PRN medication was given appropriately, nonparticipant observation could be used in a mixed-methods study.
Furthermore, there is a need for better understanding of how care home staff make decisions to administer PRN analgesics, and to further explore internal care home barriers and facilitators that may be associated with analgesic administration.

| CONCLUSIONS/RELEVANCE
Clinicians should be aware that PRN drugs may not be given as often as they anticipate (or at all) and should think carefully before prescribing only PRN analgesics. Some residents (males, older residents and residents with severe dementia) may be more at risk than others for undertreatment and regular medication should be considered.
Medical notes for PRN analgesia should include the indication to ensure that care staff have the right information available, which may increase the likelihood of residents receiving appropriate pain relief, and clinicians should review PRN use to check it is being administered appropriately.
In the UK, there are no pain management guidelines for care home residents with dementia. Policymakers should develop pragmatic guidelines specific to care home residents, including residents with different severities of dementia and those with communication difficulties, with clear guidance for the use of opioids.