Introduction
Nearing the end of the second decade of life, the human body would have reached its full potential development, up to and including cognitive functions and maturation of the central nervous system (CNS). It is from this point that it reaches the stage of inevitable decline, which continues until death (Hutchison and O’Brien, Reference Hutchison and O’Brien2007). This aging process is often accompanied by diminishing bodily and overall physiological reserve. It is worth noting that the trend and rate of functional reserve decline differs greatly between individuals, hence there are always variations in any sample population (Rockwood et al., Reference Rockwood, Song and Mitnitski2011).
Frailty – what does it mean to be “frail”?
Frailty is not an easy concept to define. Most experienced clinicians can and do correctly identify “frail” patients who come into their care, although this would be considered mostly instinctive. The challenge then lies in finding the definitions for, and quantifying the level of frailty.
The problems arise when researchers look for factors to include or exclude in an appropriate sample group (Rockwood, Reference Rockwood2005; Rockwood and Mitnitski, Reference Rockwood and Mitnitski2007). Traditionally, the focus of frailty lies in the measure of patients’ abilities in cognition, continence, mobility, and function. These are relatively simple parameters to measure and give reasonable predictive validity (Davis et al., Reference Davis, Rockwood, Mitnitski and Rockwood2011). It should be emphasized that, although there is a positive correlation between advancing age and frailty, they are not dependent variables of each other (Rockwood and Mitnitski, Reference Rockwood and Mitnitski2011). Patients who are frail are particularly predisposed to the unwanted effects of systemic anticholinergic agents. This may include delirium, xerostomia, visual disturbance, and iatrogenic impairment of cognitive function, such as short-term memory.
Some patients of advancing age may require assistance to the extent that they need multiple daily dosing of their medications put into clearly labeled slots on disposable devices. Such devices, labeled as “Days of the Week” and time slots (e.g. breakfast, lunch, and dinner time), allow patients or their carers or nursing staff in aged care facilities to administer the medication and keep track of compliance. Such systems may be known as “dosing administration aids” (DAA).
Anticholinergics – Why are they targeted?
It is established that a frail individual (at least from pharmacokinetics perspective) should be treated differently from their non-frail counterparts (Agar et al., Reference Agar, Currow, Plummer, Seidel, Carnahan and Abernethy2009). This is especially the case with medications possessing anticholinergic effects. In this section, sample medications will be used to demonstrate how anticholinergics affect the frail in a manner that differs from the effects on “normal 25-year-old Caucasian male” (Abe et al., Reference Abe, Chiba, Cirincione, Grasela, Ito and Suwa2009), in which so many phase-1 trials of medications are performed. Some of these medications may be used for their anticholinergic properties for the desired action, but these are also often associated with side effect profiles (Nishtala et al., Reference Nishtala2009).
Which medications are most frequently implicated?
Authors in this field tend to assign an artificial value as a means of quantifying the predicted anticholinergic impact of these medications (Carnahan et al., Reference Carnahan, Lund, Perry, Pollock and Culp2006). The numbers are usually between 0 and 3, with zero being “no anticholinergic activities,” and 3 being the highest single contributor to the anticholinergic burden.
The top medications (but by no means limited to) that are implicated through a review of no fewer than six articles (Tune et al., Reference Tune, Carr, Hoag and Cooper1992; Carnahan et al., Reference Carnahan, Lund, Perry, Pollock and Culp2006; Boustani et al., Reference Boustani, Campbell, Munger, Maidment and Fox2008; Minton et al., Reference Minton, Tofade and Shah2009; Fox et al., Reference Fox2011; Hori et al., Reference Hori2011), by class, in descending order of impact and frequency of prescription are as follows:
1. Tricyclic antidepressants (e.g. amitriptyline, doxepin),
2. some “typical” antipsychotics (e.g. trifluoperazine),
3. some SSRIs (e.g. paroxetine),
4. atropine*
(*not frequently used clinically, but it is often used as a benchmark for standardized scores/levels reflecting anticholinergic activities).
The cumulative anticholinergic burden considered clinically significant is 3 or greater. For example, amitriptyline on its own would satisfy that requirement, as will a combination of amantadine and atenolol, or a triple combination of frusemide, metoprolol, and digoxin (a very commonly used combination for congestive heart failure; Krum and Teerlink, Reference Krum and Teerlink2011).
What do they do to the frail patient?
One notable feature of anticholinergic “burden” (unwanted anticholinergic activity is often referred to as “burden” by the researchers in this area) is the effect on the CNS (Hori et al., Reference Hori2011). Hori et al. Reference Hori(2011) compared two groups of Alzheimer's disease patients with possible other co-morbidities who were hospitalized, and distinguished the two sample groups as “with serum anticholinergic activity (SAA) burden” and “without SAA burden.” When they compared the performance in the Mini-Mental State Examination (MMSE), for example, the groups that were free of SAA burden outperformed their counterparts by a margin of over 48% (13.16 vs. 8.89, p = 0.0367). Delirium and behavioral disturbance was also significantly differentiated between the groups, with margins sometimes as large as three-fold compared with burden-free subjects. For the purposes of this study, however, dosage adjustments were not part of the scope, and therefore no consideration was given to the dosage, method of delivery, or “doubling-up” of both in any one patient's medication regimen. In other words, any one agent will only attract a single score on the total ACB (e.g. oxybutynin given in both oral and transdermal forms will attract the score of 3 once).
Prevention and minimization
From the perspective of clinical pharmacy, the focal basis of practice appears to be one of prevention rather than “cure” (Spinewine et al., Reference Spinewine, Schmader, Barber and Hughes2007). Home Medication Review (HMR), in its current form in Australia, is based upon two basic elements – optimizing the medication regime by aligning medical practitioners’ prescribing habits with best evidence, and improving adherence by patient consultation and counseling. Both of these elements are proven to be effective in reducing hospital admission and overall cost to the public health system from a general perspective (Desborough et al., Reference Desborough2011; Maidment et al., Reference Maidment, Fox, Boustani and Katona2012). The other reason would be the quality of life for those patients. It is often associated with self-empowerment and independence, and this can only be achieved by preventing and minimizing the need for hospital admission and stay. This approach also reduces the possible compounding of the problem, as it is found, the ACB in a large number of patients actually increased after discharge from hospital (Wawruch et al., Reference Wawruch2012).
This current study takes a theoretical approach, whereby individual patients are given consideration for the possibility of a reduction in their overall ACB by means of altering their current medication regime based on relevant practice guidelines.
Method
The scope of this study was to target a group of patients who can be identified as “elderly,” and possibly frail. The population studied in this project consists of residents in sheltered accommodation or in nursing facilities with varying degrees of support for activities of daily living, and all receive their medications in a DAA. A de-identified database of 2,236 such patients was analyzed. With approval from the CSU School of Biomedical Sciences Ethics Committee, these data were drawn from a community pharmacy's DAA database, containing the profiles of medications that these patients were currently receiving, in a period of approximately four weeks. The data were extracted without identifiers by staff at Webstercare® (Leichardt, NSW) and passed to the research team as an Access™ database. Table 1 highlights some of the characteristics of the patients as given by Webstercare®.
Summary of characteristics of patients studied
Exclusions
There were some patients excluded from this analysis in order to preserve the quality of the final analysis and the impact of relevant individual results are not skewed or “diluted” in any way.
To be included, the patients needed to be using at least one regular medication with systemic effects. For this reason, any topical agents with little to no quantifiable systemic effects were excluded, such as lubricant eye drops, topical corticosteroids, and skin wash liquids. Following a similar logic, patients who are taking oral agents but with little to no quantifiable systemic effects, were also excluded. These included the likes of stand-alone daily, low-dose fiber supplements, simple antacids, and nutritional supplements (e.g. Two Cal®).
Identification
There are a few set criteria as to how and why agents are identified to be significant for the purposes of this study.
The Anticholinergic Cognitive Burden (ACB) scoring is based primarily on the works of two groups: Boustani et al. Reference Boustani, Campbell, Munger, Maidment and Fox(2008) and Sittironnarit et al. Reference Sittironnarit(2011). Boustani et al. Reference Boustani, Campbell, Munger, Maidment and Fox(2008) used a list of medications that are classified into three groups, giving each a score according to their apparent impact of anticholinergic effects on cognitive abilities. Scores are given from 1–3, with a positive correlation to the highest score with maximum impact. This forms the basis of identification of most of the agents that are included in this study. Sittironanarit et al. Reference Sittironnarit(2011) used a different and more recent list of equivalent medications using similar scales. Although the two lists are largely overlapping, they do differ in terms of the medications included and the scores assigned to them. In order to achieve an adequately comprehensive range of medications, both lists are used in this study.
In order to reduce the possibility of over-representing the extent of the impact and to achieve consistency, wherever there is discrepancy between the two groups in the scoring of individual agents, Boustani et al. Reference Boustani, Campbell, Munger, Maidment and Fox(2008) have been given precedence and adhered to throughout the study. In order to be consistent with the logic used while compiling these lists, some agents are necessarily excluded from the study. Agents that exhibit no significant systemic actions will be excluded, even if they are obviously (and intentionally) anticholinergic in nature, such as inhaled ipratropium or tiotropium. There is no definitive evidence that these agents cross the blood–brain barrier and elicit any significant impact on the CNS and therefore cognitive function. For practical considerations, short-term medications, such as a course of antibiotics or benzodiazepines, were excluded, since this study focuses on the long-term therapeutic outcomes. Similarly, only medications that were “current” in status were included, any change in dosage and/or commercial brands was counted once, since no consideration of dosage-adjustment was given for the purposes of this study.
Analysis
Given the criteria mentioned above, the number of patients was then determined to be 691 (n = 691). Each patient's profile was then analyzed, and the number and total score of implicated agents were identified using the ACB by Boustani et al. Reference Boustani, Campbell, Munger, Maidment and Fox(2008). The theoretical intervention was to attempt to identify the likely indication of each of these agents, and substitute an alternative for them wherever possible, as permissible by the current Australian Therapeutic Guidelines and Australian Medicines Handbooks as sources of evidence-based “best practice” standard (Neurology Expert Group, 2010; AMH, 2012). For example, if a patient was using amitriptyline (ACB = 3) with regular paracetamol and no antidepressants, then the likelihood of it being used as an adjuvant agent for neuropathic pain is high. However, the possibility of it being used as a sole antidepressant is also considered. In this case the results of substitution from amitriptyline to pregabalin (ACB = 0, for neuropathic pain) and to sertraline (ACB = 0, for depression) are identical, which represent an absolute reduction in total ACB by 3. Where there was a discrepancy in reduction between the two (or more) possible indications, the most conservative option was applied. An example of that would be promethazine (ACB = 3), which could be substituted with loratadine (ACB = 1, as an antihistamine) for a reduction of two points; or domperidone (ACB = 0, for nausea) for a reduction of three points, then the former is preferred unless there is obvious evidence to the contrary (e.g. the patient is concurrently using a different antihistamine).
Although this study is only based on theoretical intervention, it takes note of the limitations as to what changes are possible without impacting the overall therapeutic effectiveness of each patient's individual regime. This is a theoretical model where the prescribers’ intentions and rationale for the use of each agent is not always clear. As such, the authors did not attempt to cease any medication based on assumptions of “not needed.” Patients who are on clozapine, for example, would be largely left unchanged. That is due to the sensitive nature of the medication both in terms of side effects and drug interactions, and the fact that they are required to be under the care of a specialist psychiatrist and subjected to regular full blood examinations (FBEs) as required by practice guidelines in Australia.
The two groups of results – the initial total ACB and the final ACB after possible reductions – were then subjected to a “Paired-Samples T-test” using SPSS® in order to ascertain whether or not the two groups are “significantly different,” to demonstrate a meaningful outcome.
Results
Of the 691 patients analyzed in this study, 242 of them (35%) were found to be using at least one anticholinergic agent on a regular basis at the time of the study with significant systemic effects and thus contributing to the overall ACB. Forty-nine of these patients (20%) attracted a score of 3 or above on the ACB, which is deemed to be clinically significant (Boustani et al., Reference Boustani, Campbell, Munger, Maidment and Fox2008). Figure 1 shows the constitution of patients by total ACB. Of the 35% who are using at least one agent on the ACB scale, an absolute reduction is possible in 59% of them (n = 143).
Number of patients categorized by total (individual) AC Load.
Further, it appears possible to reduce those from having an ACB score of 3 or above to a clinically insignificant score of 2 or less in 85% of the cases (n = 42). This would mean that the total percentage of patients who are potentially affected by a clinically significant anticholinergic burden could be reduced from 7.10% to 1.01%. Figure 2 shows the number of patients classified by resultant ACB scores after theoretical interventions.
Patients grouped by final AC score for each individual studied after theoretical intervention.
In a Paired-Samples T-test, comparing the results of the ACB load before and after theoretical intervention, it was found that the correlation between the two sets of data (before and after the theoretical intervention) is 0.692 (p < 0.001). The ACB scores, however, were reduced after the intervention, and the difference is 0.331. The 95% CI (df = 690) does not contain the value of 0 (i.e. 0.272–0.391), and therefore it may be concluded that there is a significant difference between the two sets of data and the null hypothesis (H0) is rejected. This can be confirmed by the fact that significance of the t-value is <0.001.
This study also found that of the 246,960 counts of items dispensed and recorded (both prescription and non-prescription) for this group of patients, 47,334 (or 19.2%) of them were of agents on the ACB scale.
Discussion
This theoretical study found that 35% of the patients are taking at least one agent contributing to ACB, 20% of whom have a clinically significant score of 3 or above; It suggests that absolute reduction of ACB is possible in 59% of the patients, and it is possible to bring 85% of those scored 3 or more to a score of 2 or less. The exceptions to the possibility of a successful intervention can include, but are not limited to, the example of clozapine mentioned in “Methods” section. Due to the well-known potential for clozapine to cause, among other things, neutropenia or even pancytopenia, rigid monitoring of full blood count (FBC) results is mandatory for any patient enrolled into the Clozapine program (Kuluris, Reference Kuluris2007; Ronaldson et al., Reference Ronaldson, Fitzgerald, Taylor, Topliss and McNeil2011). The specific course of treatment is tailored by psychiatrists who, with expertise in their field, are also responsible for the overall therapeutic outcome of the patient. For this, and many other clinical reasons (such as drug interactions, brittle but stabilized regimens involving warfarin, phenytoin, digoxin, etc.), the physician in charge of the patient may not agree with proposed interventions from pharmacists. Therefore, at this early stage of evidence gathering, it cannot be conclusively stated that such interventions are successful or even possible. Another aspect of technical difficulties that the researchers encountered is to determine whether or not to include certain agents at various doses, since the ACB is not dosage-adjusted in the scope of this study. One significant example is codeine, which is recognized by a number of experts in the area to attract an ACB of at least 1. However, given that quantitative measures, such as SAA, are restricted to a laboratory setting and are not practicably available to a clinical study, it is difficult to ascertain whether or not lower dosage formulations should be included (Carnahan et al., Reference Carnahan, Lund, Perry, Pollock and Culp2006). Codeine phosphate 30 mg (with or without a combination with paracetamol) should obviously be considered; however, products containing doses of 15 mg or less (typically 8 mg) are less considered. That is particularly an issue when it is only used on a pro re nata (PRN) basis, whereby the frequency and dosage at each dosing time is very difficult to determine from the available data. The gross inclusion of all “codeine-based” products will skew the results to reflect that more patients are afflicted with an ACB than there are in reality – even if it only attracts a clinically insignificant score of 1 on its own. In order to keep in line with the approach for consistency and inclusiveness, the researcher has decided to include them as it is listed on the two ACB scoring lists.
Lastly, the theoretical intervention conducted in this study is limited to “ideal situations.” Many practical considerations, such as Pharmaceutical Benefits Scheme (PBS) or private cost of each medicine, are crucial in the decision-making of pharmaco-therapeutic regimens. The obvious differences, such as amitriptyline ($8.54–8.99) versus pregabalin ($42.75–183.24) in the treatment of neuropathic pain, and oxybutynin ($14.48) versus solifenacin ($56.45–79.30) for urinary incontinence, are indeed as large as these are significant (MIMS Australia, 2012). Other issues such as treatment-resistance and non-compliance are taken into account in this study, but only to the extent of “minimizing” the impact rather than complete circumvention (e.g. lowering ACB by substituting prochloperazine with domperidone instead of similarly effective and safe agent such as ondansetron for nausea – for cost reasons).
For these and many other possible reasons, further studies are absolutely necessary to validate, dispute, or otherwise build on these findings. The authors envision this study as a preliminary step to much larger, actual clinical intervention studies. At the time of writing, such intervention studies have commenced in this area. Interventions that could be suggested to the clinicians and researchers may include, for example, a protocol in which “every ACB-contributing agent intended for use is to be checked against relevant guidelines for a lower-ACB equivalent before prescribing.”
Conflict of interest
None.
Description of authors’ roles
Zikai He – chief investigator, literature review, data analysis, and report writing. Prof. Patrick Ball – supervisor, contributor of data, primary reviewer of report, and project conception and design.
Acknowledgments
We acknowledge Gerard Stevens (Managing Director) and Melissa Mirabile (Software Manager) at Manrex Pty. Ltd. Webstercare© for their contributions in the extraction and de-identification of the data; and to Lisa Wallis and William Crofts at Turvey Tops Pharmacy for their permission to allow the de-identified data to be extracted.
