2.1. Subjects and study design

Healthy male and female subjects aged ≥60 years with English as a first language (n=150) were entered into a multicentre, randomised, double-blind, double-dummy, placebo-controlled, parallel-group, 3-week study. Subjects had to be able to follow instructions and complete the computerised cognitive tests with valid responses. Medications prohibited for 2 weeks prior to screening included drugs with anticholinergic properties, drugs with known effects on cognition (e.g., opioids, benzodiazepines or sedating antihistamines), or drugs that are substrates or inhibitors of cytochrome P450 (CYP) 2D6 or CYP 3A4. Subjects were excluded if they suffered from conditions for which anticholinergic use is contraindicated (e.g., uncontrolled narrow angle glaucoma, urinary retention), if they suffered from dementia or scored ≤27 on the Mini-Mental State Exam (MMSE) or if they displayed evidence of depression (score ≥9 on Geriatric Depression Scale).

Subjects entered a 2-week screening period, during which eligibility was assessed, and cognitive tests were administered for familiarisation with procedures. Subjects were randomised (1:1:1 ratio) to receive once-daily (od) treatment with oxybutynin ER, darifenacin or placebo (Fig. 1). The 3-week duration allowed titration of oxybutynin ER in accordance with US prescribing information [16]; treatment for 1 week at each dose allowed steady-state cerebrospinal fluid concentrations to be reached. Thus, the oxybutynin ER group received 10mg od in week 1, 15mg od in week 2, and 20mg od in week 3. The darifenacin group received 7.5mg od in weeks 1 and 2 (with a sham dose increase after 1 week), then 15mg od during week 3 (in line with US prescribing information) [17]. The third group received placebo throughout, with sham dose increases after 1 and 2 weeks. Blinding was maintained by using the double-dummy technique. Dosing was supervised during week 3 to ensure compliance.

Fig. 1. Study design. CFT=cognitive function test; ER=extended release.

Written informed consent was obtained, and the study was performed in accordance with good clinical practice guidelines following ethical approval by a local review board according to the ethical principles laid down in the Declaration of Helsinki.

2.2. Assessment of cognitive function

Cognitive function was assessed through the Psychologix/CogScreen (Psychologix Inc, Fort Lauderdale, FL, USA; CogScreen, LLC, Washington, DC, USA) battery of computerised cognitive function tests (CFTs) performed during clinic visits at baseline and following each week (prior to dose or sham dose increase) (Fig. 1). Table 1 shows the tests that were employed (which have been demonstrated to be reliable and valid in numerous studies) [18], [19], tasks performed in each test and the sequence in which the tests were performed.

Table 1. Battery of tests used to assess memory and cognitive function

	Test	Description	Test sequence*
	Immediate memory recall
	Name–Face Association	Subjects are presented with a series of 14 people (displayed on a video monitor) who introduce themselves individually by common first names. Subjects are then asked to recall names when the 14 people reappear in a different sequence. Two separate tests performed (first and second acquisition).	1
	First–Last Name Association	Subjects are presented with four pairs of first and last names. Subjects are then asked to recall corresponding first names as each last name is presented. Two separate tests performed (first and second acquisition).	2
	Facial Recognition	Subjects are presented with a single facial photograph and are required to touch the face on the screen. In each of 24 subsequent trials, subjects are required to identify a new face added to the set (8-second delay between each trial). Results are analysed as ‘correct before first miss’ and ‘total correct’.	4
	Delayed memory recall
	Name–Face Association	30minutes after completion of the immediate recall Name–Face Association Test (as described above), subjects are re-presented with each face and asked for the corresponding name.	8
	First–Last Name Association	30minutes after completion of the immediate recall First–Last Name Association Test (as described above), subjects are asked to recall corresponding first names as each last name is re-presented.	9
	Misplaced Objects	Subjects are presented with a 12-room house on a monitor and asked to place 20 objects within the house using a touchscreen (no more than 2 objects per room). After a 30-minute delay subjects are asked to recall object placement. Correct recall at first attempt is assessed.	3 (presentation) 10 (delayed recall)
	Visual attention and memory
	Matching to Sample	Subjects are presented with a checkerboard pattern (4×4) made up of purple and yellow squares. The pattern disappears and is replaced by one identical and one similar pattern. Subjects are asked to identify the identical pattern. Response speed, accuracy and efficiency are measured.	5
	Visual Sequence Comparison	Subjects are presented with two random strings of numbers and letters (4–8 items) simultaneously (shown on left and right hand sides of a monitor) and are asked to identify whether they are the same or different. For each pair of strings, differences of up to two items are allowed. Speed, accuracy and efficiency (the number of problems correctly completed per minute) are measured.	6
	Psychomotor/reaction time and information processing
	Divided Attention (Visual Monitoring Alone)	Subjects watch a cursor (indicator) move vertically within a circle divided into central, upper and lower sections. When the cursor crosses into upper or lower sections, subjects are required to press a box marked ‘CENTRE’ with a light pen. Indicator speed is measured as the median time the cursor spent outside the central section of the circle before the subject presses ‘CENTRE’. Premature responses also are assessed.	7
	Divided Attention (Visual Sequence Comparison and Visual Monitoring, Dual Condition)	In the second component of the Divided Attention Test, the Visual Sequence Comparison task (as described above) is performed simultaneously with the Divided Attention Indicator Alone Task (as described above). Response speed is measured for both tasks and accuracy and efficiency (number of items completed) are measured for the Visual Sequence Comparison Task in the Dual Condition (i.e., when performed with the Divided Attention Visual Monitoring Task). When the two tasks are presented simultaneously, the test assesses divided attention, working memory, and visual-motor and visual-perceptual speed. In addition, comparison of performance under single and dual task conditions yields information regarding the subject's capacity for multitasking.	7

* Order in which tests were performed.

The primary end point was the effect of each antimuscarinic at week 3, versus placebo, on recent (delayed) memory as measured by accuracy on the delayed recall Name–Face Association Test [20]. This test measures an ability that declines markedly with advancing age [4] and has shown some limited changes in response to drugs [21]. Name-recall is the most frequent memory complaint at all ages across multiple cultures [18], [22]. This parameter is therefore relevant to the daily activities of older patients with OAB, making it appropriate for analysis. The most important secondary end points were delayed recall on the First–Last Name Association Test [19] and the Misplaced Objects Test [23], both of which measure memory abilities relevant to daily life, and on which performance declines with advancing age. Also included as secondary measures were delayed recall scores at weeks 1 and 2, and effects on immediate memory, visual attention, information processing and psychomotor/reaction time.

Subjective memory loss was assessed as a tertiary end point, using a validated self-reporting instrument, the Memory Assessment Clinics Self-Rating Scale (MAC-S) [18]. The MAC-S is a test in pencil/paper format, in which each subject is asked to rate their abilities on 10 specific memory tasks and two global items. Subjects were asked to rate how their memory had changed since the beginning of the study.

2.3. Assessment of safety and tolerability

Adverse events (AEs) (graded by severity and relationship to study drug as assessed by investigators), including serious AEs (SAEs), were documented. Results of laboratory tests and vital signs were recorded.

2.4. Statistical analyses

A sample size of 35 subjects per group was considered sufficient to detect an effect size of 0.867 for active treatment versus placebo at week 3. This decision was based on oxybutynin ER having an effect size, versus placebo, approximately one third of that seen with scopolamine in older patients [7]. Allowing for a dropout rate of 30%, enrollment continued until at least 150 subjects (≥50% female) were recruited. Thus, a 1:1:1 randomisation schedule gave approximately 50 subjects per group. Analysis of the primary end point was based on a modified intent-to-treat (ITT) population (subjects taking at least one dose of study medication with complete baseline and week 3 scores for the primary end point). For secondary end points, the modified ITT population comprised subjects with scores for at least one test at baseline and any post-baseline time point. Scores for active treatments were compared with placebo using an analysis of covariance (ANCOVA) model with baseline score, age and gender as covariates. This was a two-sided test at the 5% significance level. For exploratory purposes, comparisons between darifenacin and oxybutynin ER were derived from an identical ANCOVA model.