Addison’s disease (AD) is a rare endocrine disorder thattypically results from destruction of the adrenal cortex, and that is characterized by decreased production of glucocorticoids and mineralocorticoids. Despite replacement therapy (usually oral hydrocortisone or prednisone to replace cortisol, plus an additional mineralocorticoid (fludrocortisone) to control sodium and potassium balance), patients frequently report experiencing relatively poor quality of life (Lovas et al. 2002, 2003, 2010; Thomsen et al. 2006; Hahner et al. 2007). In particular, and aside from reportsof reduced vitality, sleep disturbances, and increased fatigue and affective problems, these patients often complain of poor memory and impaired concentration (Ten et al. 2001; Tytherleigh et al. 2004; Arlt 2009). However, few studies have provided detailed characterization, via objective testing, of cognitive function in AD.

Because of the close relationship between variations in cortisol levels and performance on tests of memory and attention/executive functioning (Newcomer et al. 1999; Lupien et al. 2008; van den Bos et al. 2009), objective assessment of AD patients’ performance in these domains is particularly pertinent. Previous studies have shown, in healthy young and older adults, that elevated cortisol levels impair cognitive functioningin ways that are predictable and that can be explained at the neurobiological level (Sapolsky et al. 1986; Lupien et al. 1999; Kim and Diamond 2002; Smeets 2011). These studies focus strictly on psychosocially or pharmacologically induced elevations in cortisol levels; in AD, cortisol levels can be elevated far above basal levels (e.g., immediately following hydrocortisone administration) or depressed far below those levels (e.g., when several hours have passed since hydrocortisone administration—the medication has a relatively short half-life of approximately 1.5 h; Harbeck et al. 2009).

Furthermore, AD patients do not exhibit the normal diurnal cortisol variation. Healthy individuals have a clear diurnal rhythm in which cortisol levels begin to rise in the early hours of the morning, surge post-awakening, and decrease steadily throughout the day (e.g., from 16 h00 to 24 h00, cortisol decreases continually until reaching a nadir that is at less than 75 % of the morning values; Krieger 1975). In contrast, AD patients have extremely low cortisol levels before their morning dose of hydrocortisone (with lower than normal levels between midnight and the early hours of the morning); after taking the medication, however, cortisol levels increase rapidly (Burke 1985). Hence, while healthy individuals release cortisol in a steady pulsatile fashion, AD patients experience fluctuating levels.

This variability could play an important role in the cognitive functioning of AD patients. The relationship between cognition and circulating glucocorticoids typically follows an inverted-U shaped pattern. Specifically, a certain level of cortisol is needed to enhance cognitive functioning; decreases below or increasesbeyond the threshold of optimal functioning impair cognition (McEwen 1997; Conrad et al. 1999). Hence, the irregular rhythm of cortisol secretion in AD patients, and its relationship to their cognitive functioning, merits investigation.

Only a few published studies have reported on objective measures of cognitive functioning in AD patients. Klement et al. (2009, 2010) found significant differences between AD patients and healthy controls on tests of selective attention and verbal memory. There are two notes of import here, however. First, these cognitive tests were administered in the course of experiments designed to test the effects of glucose administration on cognitive functioning in AD patients, and so the researchers did not administer a comprehensive battery assessing performance in a variety of cognitive domains under ordinary conditions. Second, the memory test was not one administered commonly in the clinic, and had no established psychometric properties. Third, the memory test did not require delayed recall of the word lists; therefore, it was not possible to measure how effectively patients retained information across time.

Specific aims and hypotheses

Patients with AD frequently present with memory and attention complaints, despite being on replacement therapy. Very few studies have, however, provided in-depth examination of memory and attention (and of other cognitive domains) in AD patients, comparing their performance to that of individualsfree of any chronic illness.

Hence, the present study aimed to quantify and describe the functioningof AD patients across a variety of cognitive domains, and to compare their cognitive function to that of healthy controls. Our working hypothesis was that, compared to healthy controls, AD patients would score more poorly on tests of memory and attention, but that their performance would be relatively intact on measures of other cognitive functions.

Under optimal conditions, neuropsychological assessment includes an in-person, face-to-face administration of a battery of standardized tests designed to measure functioning in various cognitive domains (Lezak et al. 2004). Such formal administration is obligatory (or, at least, highly desirable) in the context of forensic evaluations, or in particular clinical and research circumstances. In some situations, however, in-person, face-to-face administration is impractical or impracticable. For instance, in resource-limited settings, such as those one might encounter in low- or middle-income countries (LAMICs), large-scale research studies might have to use alternative means of test administration in order to increase sample size in a cost-efficient manner, and to reduce the sampling bias that might arise from excluding participants who are unable, unwilling, or not healthy enough to travel to the study site (Dura and Kiecolt-Glaser 1990; Lavrakas 1993). Using such alternative means of test administration becomes even more of an imperative when researchers in low-resource settings are studying disorders or illnesses with a low base rate of diagnosis in the population. Performing a formal, comprehensive, face-to-face neuropsychological assessment is not feasible when patients (a) are few in number, (b) are scattered geographically, and (c) suffer from a medical condition that might prevent extended testing sessions.

Telephone assessment of cognitive functioning provides a viable alternative in circumstances such as these. Advances in telecommunications technology, along with the fact that most individuals, even in LAMICs, have access to a personal handset, make telephone-based data collection increasingly attractive (Kempf and Remington 2007; Donner et al. 2011; Swanepoel and Thomas 2012). Hence, researchers have developed numerous telephone cognitive assessment batteries. Many of these serve as brief screening tools for dementia and other neurodegenerative diseases, or as a means to track changes in cognitive function longitudinally; most can detect, reliably and with good validity, the presence or absence of gross cognitive deficits (Gallo and Breitner 1995; Dombovy et al. 1998; Crooks et al. 2005; Guerini et al. 2008; Duff et al. 2009).

In the current study, we used the Brief Test of Adult Cognition by Telephone (BTACT; Tun and Lachman 2005; Lachman and Tun 2008) to assess cognitive functioning in AD patients who lived in various provinces of South Africa. The BTACT is a psychometrically sound instrument that, unlike many other telephone cognitive batteries, assesses a wide range of cognitive abilities; therefore, it can be used for multiple clinical purposes and in many different clinical populations (Tun and Lachman 2006; Gavett et al. 2013). It has, for instance, been used successfully in large-scale research measuring global cognition in young, middle-aged, and older adults (Brim et al. 2004; Gurnani et al. 2013).

Methods

Participants

This study is part of a larger research programme investigating quality of life in AD. To support that programme, patients are recruited from the South African Addison’s Disease Database (SAAD; Ross and Levitt 2011), and healthy controls are recruited using flyers and posters placed on notice boards around the university community and in the offices of large corporations in the Cape Town metropole.

This study featured a case–control design in which we selected participants to form two matched groups: 27 adult AD patients, and 27 community-dwelling volunteers who were free of any chronic illness.Footnote 1 We matched the groups on age (AD patients: M = 48.70 years, SD = 15.36, range = 20–72; controls: M = 49.04 years, SD = 15.11, range = 21–74; between-group comparison, t(52) = 0.80, p = 0.94), education (AD patients: M = 13.67 years, SD = 2.46, range = 8–17; controls: M = 13.82 years, SD = 2.62, range = 10–21; between-group comparison, t(49)Footnote 2 = 0.21, p = 0.84), sex distribution (7 males and 20 females in each group), and race distribution (24 white and 3 coloured individuals in each group).

The research ethics committees of the University of Cape Town’s Department of Psychology and Faculty of Health Sciences approved the study procedures.

Materials

Sociodemographic questionnaire

This instrument captured specific information from participants regarding (a) demographic variables (e.g., age, race), (b) their medical history, and, for AD patients, (c) type and dosage of current medication, and length of time since diagnosis.

Beck Depression Inventory-second edition (BDI-II)

AD patients frequently present with disturbances in mood, motivation, and behavior (Ten et al. 2001). Hence, we used this 21-item self-report instrument (Beck et al. 1996) to measure the intensity, severity, and depth of depression in respondents.

Brief test of adult cognition by telephone

The BTACT (Tun and Lachman 2006) is a brief (15-min) test of cognition that can be administered over the telephone. It consists of subtests that measure episodic memory, working memory, executive functioning, reasoning, and speed of processing. Although there are no test manuals or published articles detailing the formal psychometric properties (e.g., convergent and divergent validity) of the instrument, the BTACT’s developers do report that their telephonic assessment correlates well with a face-to-face administration of the same test battery (r ranging from 0.56 to 0.95), and that it possesses good test-retest reliability (r reaching up to.87).

The first BTACT subtest assesses episodic memory using the word list from the Rey Auditory-Verbal Learning Test (RAVLT; Lezak et al. 2004). Participants are read a list of 15 words and are then asked to recall as many as possible; this is the Immediate Recall trial. At the end of the assessment session, the participant is again asked to recall as many words as possible from the list; this is the Delayed Recall trial.

The rest of the BTACT subtests are administered, in the order given below, in between the episodic memory Immediate Recall and Delayed Recall trials. Working memoryis assessed using the Digit Span-Backwards subtest of the Wechsler Adult Intelligence Scale (Wechsler 1997). Participants are read increasingly longer sequences of randomly-ordered digits (from a string of two to a string of eight) and asked to repeat them back in reverse order. The test is discontinued when an examinee fails both trials at a particular sequence length.

The BTACT assesses executive functioningusing a Category Fluency task. Participants are asked to generate as many words as possible, in one minute, from a particular semantic category (Lezak et al. 2004).

The next subtest, Red-Green, assesses attention-switching/reaction time using a series of three simple two-choice response tasks. In the baseline task, participants are instructed to respond “go” when they hear the word green, and “stop” when they hear the word red. After completing 20 such trials, the examiner begins administration of the reverse baseline task. Here, the participant is instructed to make the opposite responses: “go” to red and “stop” to green. After completing 20 such trials, the examiner begins administration of the alternating task. Here, the participant is given cues as to when to use the response rule corresponding to the baseline task, and when to use that corresponding to the reverse baseline task. The examiner administers 32 such trials. The baseline tasks assess processing speed, and the alternating task assesses task-switching and inhibitory control.

The next subtest, Number Series, assesses reasoning. Participants are read five different series of five numbers each, and asked which number they think best continues each sequence (Schaie 1996). Finally, the Counting Backwards subtest assesses speed of processingby asking participants to count backwards from 100 as quickly and accurately as possible for 30 s.

Procedure

All participants provided written informed consent as part of the larger research programme in which they were enrolled. The first author (M.H.) contacted potential participants telephonically and asked whether they were willing to enrol in the study. Those who provided verbal consent were given an appointment date and time, and were told that a study representative would telephone them at that time to administer the cognitive tests. AD patients were asked when they took their morning medication; the appointment time was scheduled for 2 h later than that. By building in this 2-h time gap, we sought to ensure that testing took place after the initial effects of the medication had subsided (i.e., after the sharp peak that follows hydrocortisone administration; Groves et al. 1988; Løvås and Husebye 2007), but before cortisol levels dropped so low that cognitive performance might be affected (Rimmele et al. 2010).

At the appointed time, the first author telephoned the participant and administered the BTACT following the standard procedures detailed by Tun and Lachman (2005, 2006). She obtained verbal consent again, and assured participants of the confidentiality of their responses and their data. Participants were also told that if at any time during the test they wanted to stop they could, or if they felt uncomfortable doing a subtest they should say so and the administrator would move on to the next subtest. Furthermore, the test administrator emphasizedthat participants should not write anything down, and should do everything mentally.

We tested each AD patient and his/her matched control at the same time of day.

Data management and statistical analyses

Scoring the BTACT and deriving outcome variables

We scored the BTACT, and derived a set of outcome variables from it, using the standard procedures detailed by Tun and Lachman (2005, 2006).

Specifically, for the episodic memory subtest we scored the Immediate Recall trial by assigning one point to each word recalled correctly; we scored the Delayed Recall trial identically. We then calculated, as a measure of forgetting, the difference between the Immediate and Delayed Recall scores. We also recorded the number of false alarms (i.e., the number of times a participant produced a word that was not on the original list) on both recall trials.

For the Category Fluency subtest, the outcome was the number of unique words generated within the 1-min time limit. For the Digit Span-Backwards subtest, the outcome was the largest correct set size. For the Red-Green subtest, the outcome variables were derived from the alternating task: the number of correct responses made (out of a possible 32) and the time to complete the set of 32 trials. For the Number Series subtest, the outcome was the number of series completedcorrectly (out of a possible 5). Finally, for the Counting Backwards subtest, the outcome was the total number of numbers counted backwards in the correct order within the time limit.

Power analysis and sample selection

A power analysis suggested that the sample size be set at N = 88 (44 per group) to achieve a power of 0.75, given a medium effect size (Cohen’s d = 0.50) and an alpha of 0.05 (Erdfelder et al. 1996). However, given the rarity of AD, only 27 patients (and hence N = 54) could be enrolled. This sample size generatedstatistical power of 0.56.

Inferential analyses

We completed all analyses using SPPS version 21. We set the threshold for statistical significance (α) at 0.05, unless otherwise noted. For each of the analyses described below, we calculated the appropriate effect size estimate, and we interpreted these estimates following convention (e.g., for Cohen’s d, 0.20–0.30 = small; 0.50 = medium; > 0.80 = large).

The analyses proceeded across two stages. First, we used a series of independent-sample t-tests to compare group performance on each BTACT outcome variable. A Bonferroni correction controlled for inflated familywise error associated with multiple pairwise comparisons; in this case, the adjusted threshold for statistical significance is α = 0.05/11 = 0.005. Second, we used bivariate correlational analyses (with Pearson’s rcoefficient) to assess whether, in the AD group, disease characteristics (i.e., time since diagnosis, total hydrocortisone dose, dose/kg, and number of doses/day) influenced cognitive performance.

Results

Table 1 presents the clinical characteristics of the group of AD patients. Of note here is that three participants (all women, aged 30, 46, and 50 years, respectively) were not prescribed hydrocortisone; instead, they were prescribed prednisone. The average prednisone dose was 16.67 mg (SD = 12.58), and the average prednisone/kg ratio was 0.12 mg/kg (SD = 0.23).

Table 1 Clinical characteristics of Addison’s patients (N = 27)
Full size table

There were no significant between-group differences with regard to BDI-II scores, t(49) = −1.88, p = 0.07.

Table 2 presents between-group comparisons of BTACT performance. At the conventional level of significance, AD patients performed significantly more poorly than controls on a number of outcome variables related to episodic memory performance (number of false alarms on both the Immediate and Delayed Recall trials; amount of information lost over the delay) and to speed of processing. After the Bonferroni correction, only one between-group difference remained significant: In AD patients, the difference between the number of words produced on the Immediate and Delayed Recall trials was significantly larger than that for controls.Footnote 3

Table 2 Comparison of group performanceon the BTACT (N = 54)
Full size table

Table 3 presents the results of correlational analyses investigating associations between disease characteristics and cognitive performance. Neither total hydrocortisone dose, nor dosage/kg, nor number of doses per day correlated significantly with performance on any of the subtests. However, duration of illness correlated significantly with several indicators of episodic memory performance: number of words produced at both Immediateand Delayed Recall, and number of false alarms produced at Immediate Recall. Duration of illness also correlated significantly with performance on the Digit Span-Backwards, Category Fluency, Red-Green (number correctas well as completion time), Number Series, and Counting Backwards subtests.

Table 3 Correlation of BTACT measures with Addison’s disease clinical characteristics
Full size table

We ran a set of secondary analyses to determine whether the character of certain comorbid illnesses in the AD patient group influenced their cognitive functioning. In thepatient group, the most common comorbid illnesses were diabetes mellitus (25.9 %, seven people), hypothyroidism (22.2 %, six people), osteoporosis (11.1 %, three people), high cholesterol (11.1 %, three people), hypertension (7.4 %, two people), high blood pressure (7.4 %, two people), and Autoimmune Polysyndrome Type II (7.4 %, two people). Two separate independent samples t-test comparedBTACT performance in patients who had diabetes mellitus versus those who did not, and in patients who had hypothyroidism versus those who did not. There were no significant between-group differences on any of the outcome variables (ps > 0.06). Hence, in this limited sample, it does not appear that the character of comorbid illness has an impact on cognitive functioning in AD patients.

Discussion

We used the Brief Test of Adult Cognition by Telephone (BTACT; Tun and Lachman 2005; Lachman and Tun 2008) to assess cognitive functioning in a sample of Addison’s disease patients from South Africa. Confirming our working hypothesis, AD patients showed, relative to demographically-matched healthy controls, significantly impaired verbal declarative memory (associated with medium-to-large effect sizes). There were also statistically significant between-group differences on the BTACT speed of processing tests (associated with medium effect sizes). No other domains of cognition appeared impaired in the AD patients; hence, the working hypothesis that AD patients would show impaired attentional functioning (which would have been tapped by the working memory and executive functioning subtests of the BTACT) was not confirmed.

Regarding our statistical analyses, we used the Bonferroni correction to control for the use of multiple t-tests and to therefore avoid Type I errors (i.e., incorrectly reporting that a relationship exists between two variables). However, such corrections increase the likelihood of Type II errors (incorrectly dismissing a relationship between two variables). In public health research, it is important not to miss real effects, which would mean underestimating a real health risk (Jacobson and Jacobson 2005). Hence, the current study’s context might mean that taking the Bonferroni correction is too strict: There should be concern over making Type II, rather Type I, errors. In light of this consideration, perhaps the AD patients’ performance on the speed of processing tests should be evaluated at the conventional threshold for statistical significance (p < 0.05), and perhaps one might seek clinical significance in that finding.

Regardless of these statistical considerations around types of error, episodic memory remains clearly the domain of greatest impairment in this sample of AD patients. Profound fluctuations in glucocorticoids that underlie many of the manifestations of AD are likely to contribute to impaired memory performance. Glucocorticoids have multiple effects on the human central nervous system, but have particularly dramatic effects on the hippocampus, a brain region critical for new learning and memory (Squire 1992; Kim and Diamond 2002). More specifically, a cortisol deficiency such as that seen in AD may result in cell death in the hippocampus and prefrontal cortex (PFC), resulting in memory impairment.

However, most patients with AD are supplemented with hydrocortisone, meaning they do not experience chronic cortisol deficiencies and may even have supraphysiological levels of the hormone. Increased levels of glucocorticoids reduce hippocampal glucose uptake (De Leon et al. 1997) and neuronal excitability (Joels 2001), impair synaptic plasticity (Diamond et al. 1992; Pavlides et al. 1996), decrease the number of newly-generated neurons, and alter synaptic density in the CA1 and CA3 regions of the structure (Shors et al. 2001). Prolonged levels of increased glucocorticoids (such as continuously taking hydrocortisone medication) may produce permanent degeneration of hippocampal neurons and atrophy of dendrites in the CA3 region of the hippocampus (Sapolsky et al. 1986).

The brain-based effects of glucocorticoids are not limited to the hippocampus, of course: Cortisol release enhances dopaminergic activity and increases glutamate levels in the PFC, and also alters dendritic organization in that structure (Moghaddam 2002; Arnsten 2011). This brain structure plays an important role in declarative memory retrieval, particularly in post-retrieval monitoring processes. More specifically, during post-retrieval monitoring the PFC is involved in search and decision-making processes necessary to determine whether an event occurred in a specific context (Burgess 1996), thereby allowing the accurate reconstruction of memories.

The effects of cortisol on the hippocampus and PFC can affect memory processes because both of those regions play integral roles in memory processing, and both have dense concentrations of glucocorticoid receptors (Schacter and Wagner 1999; Alderson and Novack 2002; Kim and Diamond 2002; Shansky and Lipps 2013). Consequently, researchers hypothesize that increases in glucocorticoid levels can impair contextual and declarative memory tasks that are known to require hippocampal and PFC function (Payne et al. 1996; De Quervain et al. 2003; Arnsten 2009). These hypotheses have been confirmed by numerous studies showing that contextual and declarative memory tasksare particularly impaired by exposure to environmental stressors (Lupien et al. 1997).

The memory deficits seen in AD patients could also be explained by differential activation of brain receptors. Most of these effects of cortisol on the human hippocampus and PFC are mediated by the interaction of glucocorticoids with mineralocorticoid receptors (MRs) and glucocorticoid receptors (GRs; Wolf 2003). MRs bind naturally circulating cortisol with high affinity; they are involved in behavioural reactivity to novel situations necessary to encode new information. In contrast, GRs have a lower affinity for cortisol and only become heavily occupied after a stressor or an event that raises cortisol levels (Wolf 2003); they are involved in consolidation and retrieval of learned information (Kirschbaum et al. 1996; de Kloet et al. 1999). In support of the assertion that a balanced activation of receptors is needed for optimum memory performance, Tytherleigh et al. (2004) found that adequately treated AD patients performed significantly better on a declarative memory recall and working memory tasks when both receptor types were activated compared to when only either MRs or GRs were activated. These results suggest that a balanced activation of both receptors may be needed for optimal memory functioning in humans. Research in healthy controls has corroborated this notion. For example, de Kloet et al. (1999) showed that activation of both MRs and GRs are a prerequisite for optimal memory functioning.

The data we report here are not entirely consistent with previous research investigating cognitive function in AD patients. For instance, Klement et al. (2009, 2010) found, using the Stroop test, significantly impaired attentional functioning in their AD patients; in contrast, we found no deficits on those BTACT subtests that might tap into attentional functioning (i.e., the tests of working memory and executive functioning). In this case, the disparity in findings across studies might be attributable to the use of different (and, one might argue, deficient) measures of an exceptionally broad cognitive construct.

Klement et al. (2009, 2010) also found that their AD patients performed significantly more poorly than controls on immediate recall of a 30-item word list. In contrast, we found no significant between-group differences on the immediate recall trial of the BTACT episodic memory subtest. Of particular interest here, perhaps, is that by the delayed recall trial our AD patients had forgotten significantly more words than had the healthy controls. This pattern of data might be explained as follows: On the immediate recall trial, it is possible that patients recalled words relying solely on their working memory systems (and our results suggest that working memory was not impaired in our sample of patients). However, on the delayed recall trial, participants would need effortful retrieval strategies (requiring intact functioning of the PFC) to recall previously-learned information. As we have discussed above, abnormal cortisol levels may not allow for such efficient use of such retrieval strategies.

Duration of illness was an important predictor of performance on all BTACT subtests. Two factors might account for this pattern of association. First, because increased cortisol levels impair cognition, one might predict that a longer time since diagnosis (and therefore a longer period of time taking hydrocortisone) will result in more impaired cognitive functioning. Second, cognitive functioning is known to decline with age (Salthouse 1996; Verhaeghen and Salthouse 1997), and, in AD, a longer time since diagnosis often implies increased age. Of note is that dosage/kg was not a significant predictor of cognitive functioning. In this sample, however, the distribution of dosage/kg values had a small standard deviation, and hence there was perhaps not enough of a range to detect dosage effects on cognition.

We interpret all of our data while acknowledging that this study had several limitations. First, the relatively small sample size (a consequence of rarity of AD) meant that our statistical analyses were slightly underpowered (1 − β = 0.56). Second, although the BTACT was tolerated well and proved useful in helping us test a socioeconomically and geographically diverse group of patients, the instrument is not immune to the same criticism that faces other telephone test batteries. Specifically, well-documented limitations of telephone cognitive assessment are that (a) the impossibility of presenting visual stimuli over the telephone means that important aspects of cognitive functioning cannot be assessed, and (b) the brevity of such batteries means that they do not include sufficient items to sample the wide range of abilities typically assessed in an in-person, face-to-face neuropsychological assessment (Crooks et al. 2006; Duff et al. 2009). Third, it would be of interest to measure cortisol levels in AD patients and correlate these with cognitive functioning. By design, we assumed that when we telephoned patients they were in a physiological state where cortisol levels were neither too high nor too low (therefore providing an optimal level of functioning); we have no objective evidence to verify that assumption, however. It would also be of interest to test cognitive functioning at various time intervals after medication doses are taken to determine at what stage optimal cognitive functioning occurs in these patients. Data from such a study might prove useful in determining individual differences in what level of cortisol is needed for optimal cognitive functioning, and at what point cortisol levels begin to hinder cognition. Finally, we did not collect data on the presence of acute illness at the time of testing, or on the presence of sleep disruptions the night before testing. Such variables might have a negative impact on cognitive functioning; future studies should attempt to capture data on them, and consider their potential as confounders.

In conclusion, we showed that AD patients perform more poorly on tests of episodic memory (particularly at the delayed recall stage) than demographically-matched healthy controls. These results are not confounded by between-group differences in level of education and depression symptomatology, or by differences in the type of medication prescribed to the AD patients. We also showed that the BTACT could be used successfully to discriminate between the cognitive performance of AD patients and healthy controls. Although this latter result is promising in that it shows how successfully alternative means of assessment might be adopted in a resource-limited setting, we do acknowledge that studies featuring more comprehensive neuropsychological assessment are needed to fully explore the deficits in memory performance in AD patients. For instance, at present the literature documents only deficits in verbal declarative memory, but that might be solely because non-verbal/spatial memory has not been investigated. If alterations in cortisol secretion affect hippocampally-dependent memory, there is no reason to suspect that the effect should be material-specific; that is, there should be both verbal and spatial memory deficits (Luine et al. 1994; Schwabe et al. 2008). Furthermore, no studies haveinvestigated non-declarative or procedural forms of memory (e.g., perceptual and conceptual priming) in AD patients. Describing the full extent and nature of the memory deficits in AD is an important step toward being able to ascertain the mechanisms underlying such impairment.