Copyright is owned by the Author of the thesis.  Permission is given for 
a copy to be downloaded by an individual for the purpose of research and 
private study only.  The thesis may not be reproduced elsewhere without 
the permission of the Author. 
 



HORMONE REPLACEMENT THERAPY USE 

AND 

EVERYDAY MEMORY IN MID-AGED 

NEW ZEALAND WOMEN 

YOLANDE M HAMIL TON 

2000 



Hormone Replacement Therapy Use 
and 

Everyday Memory in Mid-Aged 
New Zealand Women 

A Thesis Presented in Partial Fulfilment 
of the Requirements for the 

Degree of Masters of Arts in Psychology 

at 

Massey University, Palmerston North, New Zealand 

Yolande M Hamilton 

2000 



ABSTRACT 

There continues to be inconsistent evidence as to the extent that Hormone 

Replacement Therapy (HRT) may preserve memory performance in menopausal 

women. The Rivermead Behavioural Memory Test-Extended Version (RBMT-E) is 

a new measure of everyday memory developed for normal populations. The 

RBMT-E was used to test the everyday memory in a sample of 104 mid-aged New 

Zealand women (53 HRT users & 51 non-users) . Measures of mood, affect, 

stress, general health and menopausal symptoms as well as age and education 

were also taken to control for possible confounds. Results showed significant 

differences (p< 0.5) between the groups for three sub-tests: 'Story Immediate', 

'Story Delayed ', and 'Message Delayed '. After calculation of a total profile score 

and adjustments for age and IQ, HRT users scored more highly than non HRT 

users on the RBMT-E overall measure of Everyday Memory. Conclusions suggest 

that HRT use does show a relationship with verbal memory, and that the potentially 

beneficial effect may assist in the performance of everyday memory tasks. 



ii 

ACKNOWLEDGEMENTS 

The completion of this thesis would not have been achieved without the support 

and encouragement of a number of people. 

To the women of Christchurch who so willingly came forward to participate in the 

study, expressing their desire to help, and giving their precious time from their ever 

busy lives, thank you . 

I would like to express my gratitude and appreciation to my supervisor, Dr. Chris 

Stephens, whose pragmatism and patience helped curb my often , impatient desire 

to be done. 

I would also acknowledge the support from the special people in Christchurch who 

unstintingly gave me their time and wisdom during the difficult times. I would like to 

express my thanks to Jim Pollard for his willingness to listen , and to Fred 

Knewstubb , who came to the rescue so often when the computer would not comply 

with my requests to deliver what I had written the day before. 

Finally, I would like to pay tribute to my Mother who passed away before the 

completion , but who was unwavering in her belief that it would be done, and to my 

Dad , who stayed the distance and walked the extra mile. 



TABLE OF CONTENTS 
Page Number 

Abstract 

Acknowledgements 

Table of Contents 

List of Tables 

List of Figures 

List of Appendices 

Chapter 

1. Introduction 

2. The Evidence of a Relationship between Memory 
and HRT use 

II 

Ill 

v 

VII 

viii 

1 

3 

Biological Mechanisms Involving Estrogen 3 

3. 

4. 

Menopause 5 

Hormone Replacement Therapy (HRT) . 6 

Memory and Menopause 8 

The Effect of HRT on Memory and Cognitive Function 9 

Summary 16 

Memory and its Measurement 

Memory 

Constructs of Memory 

Measures of Memory 

Menopause and Memory - Related Variables 

Summary 

Purpose and Formulation 

Aims of the Present Study 

Hypotheses 

Design 

18 

18 

19 

23 

25 

32 

34 

34 

35 

35 

iii 



iv 

Page Number 

5. Method 36 

Participants 36 

Measures 39 

Procedure 53 

6. Results 58 

Preparation of Data & Preliminary Analyses 58 

Hypotheses 1 63 

Hypotheses 2 65 

Hypotheses 3 65 

Hypotheses 4 68 

Hypotheses 5 70 

Results of RBMT-E Utility Assessment 73 

7. Discussion 75 

Hypotheses Testing 75 

Limitations 79 

Future Research 79 

The Utility of RBMT-E 80 

Conclusion 86 

References 87 



v 

LIST OF TABLES 

Table Page Number 

1. Demographic and Health Characteristics of HRT Use Groups 37 
and Total Sample. 

2. NART Error and Estimated IQ Classification for Total Sample. 38 

3. Personal Medical History of HRT Use Groups and Total Sample. 38 

4. Concomitant Medications Use by HRT Use Groups and Total 39 
Sample. 

5. The Variables that were significantly, differently, distributed 59 
between the HRT Use Groups, showing the Number and 
Percentage in Each Group. 

6. Digit Span Means and Standard Deviations for HRT Use Groups. 60 

7. Sub-score Means and Standard Deviations from Women 's 60 
Health Questionnaire for both HRT Use Groups 

8. Profile of Mood Scores for HRT Use Groups. 61 

9. Positive and Negative Affect Mean and Standard Deviation 62 
Scores for HRT Use Groups. 

10. Evaluation of Health , Functional Status and Wellbeing 62 
sub-scale scores from the Short-Form (SF 36) Health Survey. 

11. Means and Standard Deviations scores achieved on the 63 
Social Readjustment Rating Scale. 

12. Sub-test mean Raw Scores on the Rivermead Behavioural 64 
Memory Test - Extended Version (RBMT-E) for HRT Use Groups. 

13. RBMT -E Sub-test and Total Profile Scores: Means and 65 
Standard Deviations. 

14. Results of Hierarchichal Regression of RBMT-E Sub-test Profile 66 
Scores on Age, HRT Use. and Agehrt. 

15. Means of HRT Use Groups on RBMT-E sub-tests showing 67 
Interaction Effect between Age and HRT use. 

16. Correlations between RBMT-E and POMS and PANAS Mood 69 
sub-scale scores. 



vi 

- List of Tables continued -

17. Results of Regression of RBMT-E sub-test scores on 70 
Depression and Tension sub-scale scores of Profile 
of Moods Questionnaire. 

18. Mean scores of the HRT Use Groups on RBMT-E 71 
sub-tests when regressed on Depression and Tension 
sub-scales of the Profile of Mood Questionnaire. 

19. Comparison Mean Scores with the RBMT-E Development 73 
Study. 

20. Frequency of Maximum and Minimum Achievable Scores 74 
for RBMT-E sub-tests. 



LIST OF FIGURES 

Figure 
Page Number 

1. The form of the interaction between 'age' and 'HRT 67 
use' on RBMT-E sub-test, 'Second Names'. 

2. The form of the interaction between 'Depression' 72 
and 'HRT use' on the RBMT-E sub-test, 'Messages 
Immediate'. 

3. The form of the interaction between 'Tension ' 72 
and 'HRT use' on RBMT-E sub-test, 'Route 
Immediate'. 

vii 



LIST OF APPENDICES 

Appendix 

1. Advertisement 

2. Rivermead Behavioural Memory Test-Extended Version 

Profile of Mood Scale 

Positive and Negative Affect Scale 

Women's Health Questionnaire 

Short Form Health Survey (SF-36) 

Life Events Questionnaire 

Digit Span 

North American Reading Test (NART-2) . 

Demographics Questionnaire 

3. Information Sheet 

Consent Form 

viii 



1 

CHAPTER 1 

INTRODUCTION 

This study investigates the possible relationship between Hormone Replacement 

Therapy (HRT) and the everyday memory performance of mid-aged women . 

Building on a neuroscientific platform that has demonstrated a positive effect of 

hormone replacement on memory in animals, there is now supportive evidence of 

a similar effect on verbal memory in mid-aged menopausal women . In addition , 

more recent research has suggested that the effect of hormone replacement may 

extend to other aspects of memory. 

Menopause is a transition period that occurs in all women at mid-age, and in 

addition to the loss of natural estrogen and the cessation of periods, is associated 

with various symptoms including complaints of memory loss. Hormone 

Replacement therapy is advocated by medical professionals as a compensation for 

the lack of estrogen , and for the alleviation of accompanying symptoms. The 

medicalisation of menopause and the subsequent prescription of HRT is not 

without controversy. The inconsistent and often contradictory findings of studies 

that have endeavoured to unravel the nature of the HRT effect, have extended this 

controversy. 

Alongside this evolution has emerged a continuing debate as to how memory 

should be measured. Questions about the validity of naturalistic versus laboratory 

research have highlighted the difficulty in obtaining memory measures that 

accurately portray how an individual may perform in everyday life. With the 

suggestion that the effect of hormone replacement may extend to other aspects of 

memory, it follows that there is need to ensure that such a measure accurately 

reflects how women's memory functions in everyday experience. 



2 

The Rivermead Behavioural Memory Test is designed to be an ecologically valid 

measure of everyday memory and has proven reliability and validity. The 

Rivermead has recently been further developed to create an extended version that 

is said to allow the assessment of 'normal , younger, individuals'. The present 

study utilises the Rivermead Behavioural Memory Test-Extended Version , to 

investigate the relationship of HRT with everyday memory performance in mid­

aged New Zealand women . 

Chapter 2 of this thesis examines the evidence for a relationship between memory 

and HRT. Chapter 3 describes the current understanding of the constructs of 

memory, including the concept of everyday memory. Next, how memory has been 

measured is discussed. The relationship of memory to menopause follows, with a 

description of the many confounding variables such as age, education , health , 

sleep , stress and mood that have often made it difficult to clarify the role HRT may 

play in memory performance. Chapter 4 describes the formulation and design of 

the project and outlines the specific hypotheses tested . 



CHAPTER 2 

THE EVIDENCE FOR A RELATIONSHIP BETWEEN MEMORY 
AND HORMONE REPLACEMENT THERAPY 

3 

The review of the evidence for a relationship between memory and hormone 

replacement therapy commences with support from the field of neuroscience. 

Highlights of recent findings from this arena have set the stage for past, and 

present investigations of whether memory is maintained by estrogen . Our current 

understanding of menopause, and the hormonal changes that accompany it are 

described next. A review of the research regarding hormone replacement therapy 

(HRT) follows and this includes evidence for the benefits of HRT, (those well 

established and those still inconsistent and contradictory) , as well as the 

recognised risks and contraindications for particular women. Over the past 

decade, research has endeavoured to establish that the estrogen loss associated 

with menopause does indeed impair aspects of memory. In addition , research 

findings that HRT users often exhibit a better performance in memory function than 

HRT non-users is discussed. 

BIOLOGICAL MECHANISMS INVOLVING ESTROGEN 

In recent years there has been an increasing interest in the role the central nervous 

system (CNS) plays in the control of menopause, and in turn , what effects the 

hormone estrogen may exert on the CNS during this time (Richards, Kuh , Hardy & 

Wadsworth, 1999). Estrogen appears to have an organisational and activational 

influence affecting the development and function of the CNS, and may also be 

responsible for the sexual differentiation of tissues in specific areas of the brain 

(Sherwin, 1998). 



4 

Animal studies have shown multiple sites of estrogen action, some of which play 

important roles in learning and memory function. These include the cerebral 

cortex, hypothalamus, amygdala, thalamus, the preoptic area , anterior pituitary, 

basal forebrain and the CA 1 region of the hippocampus (Woolley & McEwen, 1993; 

McEwen, Alves , Bulloch & Weiland , 1997). 

Estrogen is also known to influence several neurotransmitter systems, including 

acetycholine, noradrenaline, serotonin and dopamine (Henderson , 1997). 

Estrogen enhances cholinergic function that is known to be deficient in Alzheimer's 

disease by increasing the enzyme acetyltransferase in several areas of the brain , 

including the CA 1 area of the hippocampus (Rice, Graves , Mccurry & Larsen , 

1997). 

The neuronal atrophy that occurs in both the aging rat and human brain is 

expressed predominately in these cholinergic-rich regions of the CNS, including 

the basal forebrain , hippocampus and amygdala (Birge, 1996). Estrogen is 

thought to be responsible for stimulating neuronal repair, assisting in neuronal 

suNival and function through neurotrophic growth factors (NGF) (Birge, 1996). For 

example, adult rats , at 28 weeks after ovariectomy, have been shown to perform 

poorly on tasks of memory and learning , as well as exhibiting a decline in 

cholinergic neuronal activity in both frontal cortex and hippocampal regions. 

This deterioration appears to be prevented by the administration of estrogen 

replacement. It has been suggested that certain neuronal elements of the CNS (in 

particular those affected by memory loss disorders such as Alzheimers) are likely 

to be dependent on estrogen for their survival and function (Birge, 1996). The 

stimulation of neuronal regeneration and neurotropic growth factors evidenced by 

estrogen replacement has been shown to enhance learning (Birge, 1996). 

Estrogen effects are also suggested in the mitigation of neurotoxic effects of 

stress-induced release of corticosteroids (Henderson, 1997). An increase in 



Glucocorticoid Hormones appear to be able to damage hipppocampal neurl 

(Luine, 1994) and may be an additional mechanism through which estrogen 

influences hippocampal aging (Nappi et al. , 1999). This has important implications 

in furthering our understanding of the effects of stress on memory performance. 

Sherwin (1998), has rightly placed an important caveat on the assumption that 

changes in rats immediately apply to humans. The application of such findings 

requires replication or at least validation with human populations. With this caveat 

in mind , the findings from neuroscience described above clearly illustrate that 

estrogen, in addition to playing an important role in both the timing and 

organisation of menopause, may also exert a far wider effect on a variety of 

mechanisms including the preservation of cognitive function . 

MENOPAUSE 

The menopause is recognised as a universal event in women's life cycle . 

Menopause has been considered a process of many months or even years 

(Stevens-Long & Common, 1992), but is now more universally defined as the 

cessation of menses for longer than six months (Pearlstein, 1995). In the western 

world , menopause occurs at the average age of 50.8 years, and marks a period of 

transition for women as they move from a reproductive to a non-reproductive 

phase of the life cycle (Sherwin , 1994). 

Menopause is thought to be a multifactorial process involving both neural and 

ovarian factors. At menopause, as well as age-related alterations in hypothalamic 

function , the ovarian follicle supply is finally exhausted leading to a decrease in the 

production of estrogen (Sherwin , 1994). 

Prior to the approximate age of 40 years during the premenopausal phase, the 

ovaries are responsible for the secretion of 95 per cent of the estrogen that enters 



6 

the circulation (Sherwin , 1994). After this age, due to a continued diminishing 

number of oocytes, the ovaries produce decreasing amounts of estrogen, until 

finally there is insufficient hormone to sustain the monthly menstrual cycle. At this 

stage, the ovary no longer secretes estrogen and a dramatic decrease in plasma 

levels can be seen (Phillips & Sherwin, 1992). At this point, estrone, an estrogen 

that is a much weaker steroid , and that arises from a peripheral conversion from 

androstenedione becomes the predominant estrogen (Sherwin , 1994 ). 

Certain physical symptoms are common during this period . These include hot 

flushes , night sweats and vaginal dryness. Complaints of memory problems 

together with an inability to concentrate are also among symptoms that have been 

associated with menopause (Phillips et al. , 1992). 

There is some evidence that suggests that menopause may be a fairly benign 

event for most women , and also that the transition can often represent an active 

reorganisation of life goals and attitudes (Slaven & Lee, 1998). However, from a 

biological perspective, there is increasing support for the implication of estrogen 

loss in being responsible for a wider decremental effect on the cognitive functioning 

of women in old age (McEwen et al. , 1997). 

HORMONE REPLACEMENT THERAPY (HRT) 

The decline of the natural estrogens during the menopausal transition and the 

appearance of associated symptoms has led to the development and use of HRT 

to compensate for this decline. Prior to the onset of the menopause, three major 

types of natural female estrogens are produced . Estradiol is the more potent form , 

which is primarily produced during reproductive years. During menopause, estrone 

(a less active estrogen) becomes the more dominant estrogen and is produced by 

the conversion of androgens in adipose tissue. Estriol is the third and weakest of 

the natural estrogens (Coney, 1991 ). 



7 

The culmination of the increased knowledge about the decline in the amount of 

natural estrogens, and a greater awareness of the disruptive nature of symptoms, 

resulted in a medicalisation of menopause that introduced a theory of estrogen 

deficiency (Coney, 1991 ). As a result , efforts to limit both the menopausal 

symptoms, and an increased incidence of associated diseases, have contributed to 

the design of hormone replacement therapy (Coney, 1991 ; Mayeaux & Johnson, 

1996). 

There are three basic HRT regimens: Hormone Replacement Therapy (HRT) is 

defined as a combination of estrogen and progesterone therapy that is prescribed 

for women who show evidence of, and seek relief for, menopausal symptoms. 

This regimen can involve the estrogen and progesterone supplied simultaneously 

or sequentially (Coney, 1991 ). Estrogen replacement therapy (ERT) is a term used 

to describe estrogen therapy alone, or unopposed estrogen. This treatment is now 

generally reserved for those women who no longer have their uterus, as it has 

been strongly linked to endometrial hyperplasia and uterine cancer (Mayeaux et 

al. , 1996). For the purpose of this thesis , the definition HRT will include both 

treatments unless indicated otherwise. 

To date, the primary reason for the prescription of HRT for mid-aged women has 

been to alleviate a wide range of reported possible symptoms (Consensus 

Development Conference, 1993). The determination of which women are actually 

prescribed HRT depends to a large extent on the severity of the symptoms, the 

women's acceptance of their doctor's advice, and the amount of personal 

knowledge about HRT. The acknowledged risks of HRT use include endometrial 

and breast cancer and regular screening for these risks is recommended (Mayeaux 

et al., 1996). 

HRT has been developed from estrogens that were first derived from pregnant 

mares' urine and this continues to be the major source. In addition to oral tablet 

form, administration of HRT can also occur by the application of transdermal 



8 

patches, subcutaneous implants and intravaginal creams, although the latter does 

not appear to have a major effect in menopausal symptom relief except in large or 

long-term doses (Mayeaux et al., 1996). 

Following its development and inception , HRT has proved to be a valuable 

treatment in the reduction of long term disease (e.g. , osteoporosis , cardiac events 

and mortality outcomes) (Newman, 1999). Together with the support of 

neuroscience and empirical studies, evidence has also emerged that suggests that 

HRT use may also be implicated in the maintenance and possible enhancement of 

memory (Sherwin , 1998), and as a protective factor against various types of 

dementias (e.g. , Alzheimers , Vascular type) (Birge, 1997). 

Although these recent findings may sharpen the focus of research, the 

inconsistency of much of the evidence serves notice that the issues of 'if' and 'how' 

HRT acts is at present unclear. These issues have become increasingly complex 

to investigate, with definite conclusions remaining elusive. However, the 

importance of such a determination can not be understated . If HRT use can and 

does exhibit enhancement and protective factors to older women , it is important to 

investigate the 'face validity' of such claims and demonstrate clearly what the 

practical outcomes of HRT use for menopausal women may be. This is essential 

in order to allow women a fully informed choice as to how HRT use may benefit 

particular individual women. 

MEMORY AND MENOPAUSE 

The relationship of cognitive and memory decline to changes that occur during the 

menopausal transition, has been largely based on the evidence of behavioural 

studies in animals. These studies have established a correlation of changes in 

neural architecture with hormone levels (McEwen et al., 1997). However, although 

there is also indirect evidence that estrogen has positive effects on cognitive 



9 

function in postmenopausal women (Sherwin , 1994) , data on the effect of estrogen 

loss on cognitive performance in untreated women are few. It does not necessarily 

follow that the loss of ovarian function and the correspond ing reduction in estrogen 

will result in a decline in memory function in all menopausal women . There is 

evidence that some women experience little or any decline at all in memory 

function (Birge, 1996). Furthermore, a recent study of both male and female mid­

aged participants, failed to find any gender differences in the decline of cognitive 

function (Barrett-Connor & Kritz-Silverstein , 1999). This suggests that the estrogen 

deficiency associated with menopause may not be strongly associated with 

memory losses reported . 

On the other hand , memory loss does appear to be a prominent compla int of 

postmenopausal women (Anderson , Hamburger, Liu , & Rebar, 1987). In support 

of the association of menopause and memory loss, a more recent study (Nappi et 

al. , 1999) found that both surgical menopause and physiological menopause 

exerted a negative effect on short-term memory, whereas long-term memory, 

attention and psychomotor performances were largely unaffected . In addition , 

surgical menopause appeared to produce a more negative effect on short-term 

verbal memory than physiological menopause. 

The EFFECT of HRT on MEMORY and COGNITIVE FUNCTIONING 

In recent years there have been substantial reviews of research that has 

investigated the effect of HRT on women's cognition at menopause and in later life 

(Erkkola , 1996; Haskell , Richardson & Howitz, 1997; Rice et al. , 1997; Sherwin , 

1997; 1998; Yaffe, Sawaya, Lierberburg & Grady, 1998; Newman, 1999). While 

supporting the plausibility of the biological mechanisms that suggest an estrogen 

effect on memory, and noting the data that provides evidence of a positive 

relationship between the two (Rice et al. , 1997; Yaffe et al. , 1998), conclusions 

remain imprecise. Criticisms have included: 



10 

"Insufficiently investigated , poorly investigated" (Erkola et al., 1996, S30); 

"not sufficient evidence to recommend widespread use" (Haskell et al., 

1997, p.1251 ; Newman, 1999, p.1268); "inadequate evidence, trials too 

inadequate to recommend for treatment of dementing disorders" (Yaffe et 

al. , 1998, p.694); "evidence remains controversial with a need for more data 

and further elucidation of specific cognitive domains that may respond to 

estrogen" (Rice et al., 1997, p. 26S) . 

On the other hand , Sherwin (1998) , suggests that despite the methodological 

problems that have impinged on the conclusions drawn, there is an "overwhelming 

finding that estrogen serves to maintain or enhance verbal memory whereas it has 

little effect on visual or spatial memory in women" (p.21 ). Of particular note and 

interest, is the fact that the women who received placebo treatment in these 

studies (Sherwin , 1988; 1990) did in fact complain more spontaneously of memory 

deficit, than those women who were using HRT. 

Research investigating any possible effects of estrogen on memory and cognition 

in mid aged women has used both experimental and observational studies. 

Observational studies have in general focused on the association of estrogen with 

performance on cognitive and/or memory tests (Newman, 1999). These studies 

have mostly used cross sectional data and compared HRT current users with non 

HRT users. As a result, findings have been somewhat inconsistent. 

After matching for age and education , Kampen & Sherwin (1994) found significant 

improvement only on one test, paragraph recall. Similarly, Robinson, Freidman, 

Marcus, Tinklenberg & Yesavage (1994), found significant improvement in HRT 

users for name recall, but not for word recall. Both of these studies involved post 

menopausal women and excluded those who were currently being treated for 

depression. Kimura (1995) employed 10 cognitive tests and a test of mood and 

subsequently described "overall" improvement. However, formal adjustment for 



11 

age and years of education was not undertaken and specific test results not 

reported. 

More recently, Schmidt et al. (1996) , used a cross-sectional design to determine 

any beneficial effect of estrogen on demanding cognitive tests , and to investigate if 

this benefit was due to the prevention of silent ischemic brain damage. A positive 

association was found between HRT and cognitive functioning and a lower rate of 

clinically unsuspected ischemic brain damage in post-menopausal women. Similar 

to previous studies, this study was limited by a relatively small sample size (70 

participants using HRT) , and may not have been truly representative of the entire 

population despite the use of random selection. The mean age of the women was 

59 years and was somewhat younger than the previously mentioned studies. 

In contrast to observational studies that have illustrated improvement only in 

paragraph recall and proper name recall, Schmidt et al. (1996) found evidence of 

improvement on tasks measuring complex problem solving and psychomotor 

speed as well as visuospatial and verbal memory. No evidence was found for 

group difference on mood. In a more recent cross-sectional observational study, 

Steffens et al. (1999), investigated the association between the history of estrogen 

use and cognitive function in a large sample of community-dwelling older women 

(65 years and older). However, no specific measures of memory were undertaken 

in this study. Cognitive assessment in this study consisted only of the 

administration of the modified Mini-Mental State Examination (3 MMSE), and did 

not allow identification of what specific aspects of cognition may have been 

influenced by estrogen. In addition , the scores for all user groups were all in the 

normal range which may indicate that a ceiling effect was reached (Newman, 

1999). Although the effect of estrogen was reported as a positive one, the effect 

was quite small and it remains difficult to discern whether this effect was really due 

to estrogen use or some other design factor (Newman, 1999). 



12 

In a later, longitudinal study with a larger sample than most observational studies 

(727 participants), Jacobs et al. (1998), evaluated the relationship between a 

history of estrogen use and cognitive test performance. Measures of cognitive 

functioning here included standardised tests of memory, language, and abstract 

reasoning. Similar to other studies, HRT users were found to score higher at 

baseline, but in addition , scores on verbal memory improved over time. This study 

is important in that it suggests that estrogen benefits may extend to more global 

cognitive domains such as abstract reasoning and language as well . 

In contrast to the positive findings of the above observational studies, there are 

others that have reported an apparent nil effect of estrogen on memory and 

cognitive function in women (Barrett-Connor & Kritz-Silverstein , 1993, 1999; 

Matthews, Cauley, Yaffe, & Zmuda, 1999). Two of these studies have utilised 

large samples and appear methodologically sound (Barrett-Connor et al. , 1993; 

Matthews et al. , 1999). Matthews et al. (1999) , evaluated a cross-sectional and 

longitudinal association of HRT and cognitive function . Education , next to age was 

found to be the more powerful predictor of cognitive function than HRT use. Past 

and current estrogen users did not consistently perform better on cognitive tests. 

However, there were no specific tests of verbal memory, and because of drop out, 

longitudinal data only included the better functioning group. This study also 

demonstrated the possibility that a selection bias may be operating where 

education and HRT have been shown to be associated with higher cognitive 

function . 

The Jacobs et al. (1998) and Barrett et al. (1993) studies are similar in that both 

employed large samples and standardised measures, although the tests used were 

different. It is possible that the contrasting results could reflect cohort differences. 

Barrett-Connor et al. (1993) reported data from a highly educated and exclusive 

sample, whereas Jacobs et al. (1998) was ethnically diverse, but of limited 

educational attainment. 



13 

In a more recent study, Barrett-Connor et al. (1999) , used a cross-sectional study 

of men and women to examine gender differences specifically with memory. This 

study hypothesised that if estrogen deficiency is associated with memory loss in 

post-menopausal women , men should have less memory loss with age than 

women . However, similar patterns of decline were found with men after 

adjustments for age, depression and estrogen use. The authors conclude that the 

absence of sex differences in decline indicates a lack of association between 

estrogen deficiency and a decline in cognitive function . This finding is important as 

this is the first study that has attempted to determine whether there are sex 

differences in cognitive functions in age matched groups of elderly men and 

women . 

The Barrett-Connor (1999) study supports the suggestion that the aging process 

itself may account for cognitive decline and so act independent of any hormonal 

effect (Sherwin, 1997). Further research is needed in the future to elucidate the 

individual contributions of the aging process and hormones and areas of possible 

interaction . 

A consistent trend in the research has reported positive benefits of estrogen use, 

and furthermore suggest that the benefit may be specific to verbal memory 

(Sherwin, 1988; Sherwin et al ., 1992). However, this finding may be due to a lack 

of specific measures of visuospatial memory, or insufficient power. A more recent 

cross-sectional , observational study found that women receiving HRT had fewer 

errors on a specific measure of short term visual memory, visual perception and 

constructional skills (Benton Visual Retention Test - BVRT) (Resnick, Metter & 

Zonderman, 1997). A longitudinal analysis in this same study also showed that 

HRT users maintained their performance over time, whereas non HRT users 

showed the typical age related increase in memory errors. 

This finding has been further supported by a study that utilised positron emission 

tomography (PET), and neuropsychological assessments to examine brain 



14 

structure and function in individuals 55 years and older (Resnick, Maki , Golski , 

Kraut & Zonderman, 1998). HRT users showed better performance on 

neuropsychological tests of verbal and figural memory, and also showed significant 

differences in PET activation patterns during memory tasks. The findings of this 

study are not only credible from the point of their use of sophisticated imaging 

techniques , but also for the substantial battery of valid, reliable neuropsychological 

tests employed . Five cognitive domains were assessed : verbal knowledge, 

language, learning and memory, visuospatial abilities and perceptual speed . In 

addition , three items from the Rivermead Behavioural Memory Test were used to 

measure prospective memory. Unfortunately, data specific to this test were not 

reported . The major drawback of the study is that it is observational , and again, as 

in previous studies, differences may reflect baseline differences in the 

characteristics of women who choose to receive H RT. 

Randomised , controlled trials have also provided empirical data to analyse the 

effects of HRT on cognition and serve to control issues of bias and confound (Rice 

et al. , 1997). On the whole , experimental studies have found a positive 

relationship between estrogen use and cognition and more specifically verbal 

memory in both post surgical and natural menopausal women (Sherwin et al. , 

1988, 1990, 1992; Fedor-Freybergh , 1977; Wolf et al. , 1999). 

However, as with the observational studies, there are studies of equal merit that 

provide contradictory data (Ditkoff, Crary, Cristo, & Lobo, 1991 ; Vanhulle & Demo, 

1976; Polo-Kantola, Partin , Helenius, lrjala & Erkkola, 1998). In both the 

observational and experimental studies, methodological problems are recognised 

as being responsible for the inconsistencies (Sherwin , 1992). Sample sizes have 

typically been small . In addition , samples are often obtained from other surveys 

(e.g. , Barrett-Connor et al. , 1993) and may include the confounding effects of 

underlying disease or medications that may counter any positive effect of estrogen 

(Polo-Kantola et al., 1998). 



15 

As mentioned above, various types of tests that measure different aspects of 

cognitive functioning have been employed which make comparisons of results 

difficult. Validity and reliability of these tests is not always available (Sherwin , 

1998). Also tests may tap specific functions of memory that are not affected by 

estrogen levels (Sherwin , 1998). 

A further methodological difficulty that has been evident in previous studies is that 

different types and differing doses of estrogen preparations have been used, and 

circulating hormones not always measured (Sherwin , 1997). The path to a more 

definitive explanation for how HRT may act to preserve memory funct ion has not 

always been clear. This is due in large part to the different regimens of HRT 

prescribed , and the differing effects these may have on individual women (Sherwin , 

1998). Recently, the particular dose, type and route of administration of estrogen 

preparations has been under scrutiny (Steffens et al. , 1999). It is thought that 

these differences may alter the impact that estrogen may have on memory 

functioning . This degree of impact is likely due to a differential absorption and 

metabolisation by the liver, as well as the degree of diffusion of the different 

estrogens into the brain (Sherwin , 1997). 

A dose response relationship of estradiol to memory enhancement has been 

observed in plasma levels. There is also evidence that the brain remains sensitive 

to estrogens after a considerable period of low plasma estradiol concentrations 

(Wolf, Kudielka, Hellhammer, Torber, McEwen & Kirschbaum 1999). However, in 

contrast, Steffens et al. (1999) found no evidence of a dose-response relationship 

when measuring Mini-Mental Status Exam Scores (3MMSE). 

Despite the sharpened focus that the measuring of estrogen levels bring to the 

investigation of an HRT effect on cognition , the findings remain inconsistent. Also , 

the demonstrated effect of estrogen on verbal memory to date appears to be 

modest. Although the small degree of 'better performance' on a number of 

demanding neuropsychological tests may indicate a positive effect of estrogen , the 



16 

clinical relevance of these findings has yet to be addressed (Sherwin , 1998). 

Replicated data does not necessarily mean that untreated surgically menopausal 

women may be clinically impaired in a way that might affect their daily functioning 

in the real world (Sherwin, 1998). 

Despite this conundrum, menopausal and post-menopausal women do 

spontaneously complain of memory deficits and to date, research has not 

examined what impact HRT use may have those on aspects of memory utilised in 

the performance of everyday tasks. Furthermore, the ecological validity of the 

types of tests used to assess the effect of HRT on memory performance has not 

been established. As a consequence, there remains a need to examine the 

relationship between HRT use and memory ability in everyday functioning. 

SUMMARY 

There is now accumulated evidence from neuroscience that provides an empirical 

basis for the view that estrogen exerts effects in areas of the CNS that are known 

to be involved with the production of memory and learning (Birge, 1997). Estrogen 

mechanisms also appear to be involved in the mediation of NGF and 

neurotransmitters that are known to be associated with memory pathology. In 

addition , animal studies have shown that lowered levels of circulating estrogen do 

contribute to neuronal loss and also, that these changes can be reversed by the 

administration of estrogen (Luine, 1994; McEwen et al., 1997). This finding further 

contributes to the evidence that estrogen loss may affect memory. However, the 

generalisation of these observed changes in rats to the possibility of similar 

changes occurring in humans needs to be made with care (Sherwin, 1998). 

Menopause in human beings is recognised as a universal event, a process that 

involves neural and ovarian factors and is associated with a corresponding 

decrease in estrogen . Currently, HRT is prescribed to compensate for this loss of 

estrogen and to assist in the alleviation of symptoms. Although estrogen loss at 



17 

menopause is well documented and circulating estrogen levels have been shown 

to be associated with memory function (Wolf et al ., 1999), studies that have 

investigated the effect of HRT on memory continue to produce contradictory and in 

some part, unconvincing evidence. This is possibly due to methodological 

problems including different measures of memory, small samples and a bias in the 

self selected nature of the sample. Modest evidence for a positive association of 

HRT use with verbal memory and more recently, visual and more complex 

cognitive functioning has been shown. As yet though , there is not enough 

evidence to recommend HRT for enhancement of memory in older women . In 

addition , the clinical relevance of such evidence is unclear. This lack of clarity is 

due in part to a lack of ecological validity in the research to date. 



18 

CHAPTER 3 

MEMORY AND ITS MEASUREMENT 

MEMORY 

Memory is just one component of an individual's cognitive abilities which arise from 

a complex central nervous system (Sherwin , 1998). Memory itself, is also not a 

unitary system , but composed of processes involved in the reg istration , storage 

and retrieval of information acquired through the senses (Lezak, 1995). 

Everyday Memory has been defined as: 

"memory that is involved in the performance of everyday life tasks and 

measured either outside of the laboratory, or using simulated everyday life 

tasks , inside the laboratory" (Tomer, Larrabee & Crook, 1994, p.606). 

The term 'everyday memory' originated from an address by Neisser (1978) that 

sharply criticised the sterility of memory research undertaken in the laboratory, 

arguing that it failed to capture , or to represent memory as it occurred in the real 

world . A spirited discussion concerning issues of generalisability and theoretical 

worth followed this address (Banjai & Crowder, 1989; Ceci & Bronfenner, 1991 ). 

What has emerged out of an ongoing debate since that address, is a wide 

acceptance that it is possible, and helpful to study memory outside the laboratory. 

However, there is acknowledgement that issues of control (e.g. , material to be 

remembered , circumstances of presentation and participant's motivation) , as well 

as issues of reliability and validity remain problematic (Gathercole & Collins, 1992; 

de Wall , Wilson & Baddeley, 1994). With the development of the 'everyday' 

approach , the study of memory has incorporated memory events that more closely 

correspond to the 'everyday' memory tasks and can be considered more 

ecologically valid. 

. .. _ 



19 

Methods used to assess everyday memory have included natural observations, 

simulations of work activities and self-report questionnaires (Pollina, Greene, 

Tunick & Puckett, 1992). More recently, attempts to study everyday memory have 

developed an approach that resembles a bridge between laboratory assessment 

and assessment obtained by questionnaire and observation (e.g., The Rivermead 

Behavioural Memory Test; Wilson , Cockburn , Baddeley & Hiorns, 1989). 

Aside from the controversy mentioned above, the concept of everyday memory at 

present seems to embrace and reflect specific aspects of memory (de Wall , Wilson 

& Baddeley, 1994) that have been established in the laboratory. Those aspects of 

memory that have been identified as being implicated in the impaired performance 

or failure of everyday tasks include short-term memory (immediate or delayed 

recall) , visual and non verbal memory, recall and recognition memory, and 

prospective memory. How these aspects of memory are measured (i.e., the tasks 

employed) , and the factors that are known to affect the memory are now described. 

CONSTRUCTS OF MEMORY 

Short-term memory (STM) deals with memory that has just been presented and is 

still in the individual's conscious awareness. STM tasks that involve holding the 

material passively have been defined as primary memory, whereas the 

manipulation of information while being held is referred to as working memory 

(Craik, Anderson, Kerr & Li , 1995). STM decays in milliseconds, is attention 

dependent and can be stored from approximately 30 seconds to one hour (Lezak, 

1995). The capacity of STM is recognised as equivalent to the number of items 

that a person can hold in mind at the same time. This is usually measured by a 

"span", such as the longest number of digits or words that can be reproduced 

accurately. 

Working memory has been described as a system that allows the temporary 

holding and manipulation of information while other cognitive tasks are performed 



20 

(Baddeley, 1997). Theoretical components of working memory are thought to be 

· responsible for the strategies that are employed to manage incoming information , 

and for the maintenance and alteration of attention (Baddeley, 1997). Selective, 

sustained and divided attention are thought to be mediated by working memory 

and it appears difficulties in these aspects of attention have been identified as early 

occurrences in Alzheimer's-type dementia (Glass, 1999). 

In addition to the distinction of STM, there are memories that appear to be specific 

to the nature of the information to be learned (e.g., verbal and non-verbal 

information) (Lezak, 1995). Verbal memory tasks utilise tasks that measure the 

length of digit span (i.e., the immediate recall of numbers) , word list recall , and 

story recall (immediate and delayed) (Lezak, 1995). 

Long-term memory (L TM) refers to the ability to store information and involves 

tasks that require memory for material that has left conscious awareness from 

anything from seconds to years ago (Lezak, 1995; Craik & Jennings, 1992). As 

described above, LTM is therefore defined as by default (i.e ., anything that is not 

currently in consciousness) , and is made apparent by the ability to recall 

information after a delayed interval (Glass, 1999). However, it may not always be 

possible to clearly establish whether information presented 20 minutes earlier and 

subjected to a delayed recall is actually from L TM (Groeger, 1997). Theories have 

also established distinctive components of L TM, including episodic (involving the 

encoding and retrieval of specific autobiographical events) ; semantic (context 

independent) and more recently implicit (memory requiring the performance of a 

task) and explicit memory (memory for information is required) (Craik et al. , 1995). 

Non-verbal or Visual Memory can be defined as memory for pictures, faces and 

spatial information (Craik et al., 1995). Visual memory tasks include face, picture, 

line drawing or object recognition and recall (Lezak, 1995). Another type of non­

verbal information is the recall of personal behaviours that may be performed on an 

everyday basis. Tasks include performing a sequence of activities such as 



21 

retracing a route around a room and observing retention levels (Hess & Pullen, 

1996; Wilson et al. , 1999). Furthermore, non verbal spatial information that allows 

the navigation around , and the location of objects in an individual's environment, is 

assessed by tasks that test location recall of objects that are visually presented on 

display. Familiarity and the use of effective strategies appear to affect the 

performance of these tasks (Hess et al ., 1996). 

The effectiveness of memory is dependent on retrieval which involves both 

recognition and recall aspects (Lezak, 1995). Recognition memory occurs when a 

similar stimulus triggers awareness. Assessment involves either the individual 

making a decision about a previous occurrence, or distinguishing between items 

previously encountered (Lezak, 1995). Most studies assess accuracy in a way that 

takes into account both accurate and inaccurate (i .e., false negatives; Hess et al ., 

1996). Tasks include recognition of faces , pictures or words such as the 

Warrington Recognition Memory Test (1984)(Lezak, 1995) and can be cued or 

uncued. Faces are often paired with a name which is later used as a cue for recall. 

In contrast, recall memory occurs when information is retrieved as a result of an 

active complex search process (e.g., "Who is the President of the USA?" ; Lezak, 

1995). Recall involves 'free memory' and is assessed by using only a general cue 

(Banich , 1997). However, assessment of recall can also involve the use of direct 

cues, for example, "One of the photos had something to do with cooking - Can you 

remember what it depicted?" Recall memory is generally thought to be more 

difficult than recognition memory and involve more effortful processing (Craik & 

McDowd, 1987). 

Apart from aging , factors that affect recall and recognition appear to be the 

strength of the cue, familiarity, and context (Hess et al ., 1996). In addition , 

although the evidence is not clear, depression has been associated with deficits in 

both recall and recognition (Baddeley, 1990). 



22 

Prospective Memory is defined as the ability to remember to perform activities in 

the future (e.g ., remembering to keep an appointment or to give someone a 

message) and is in contrast to retrospective memory which is memory for past 

events (remembering someone's name) (Einstein & McDaniel , 1990). Data 

collected from both naturalistic and laboratory settings (Cherry & Lecompte, 1999) 

have identified a number of factors that affect impairment in prospective memory 

tasks. Data from both settings have also highlighted specific age-related paradox 

and difficulties with the research of this memory task (Rendell & Thomson , 1999). 

Typical procedures for testing prospective memory have involved asking subjects 

to perform some action in naturalistic settings at specified times in the future (e.g. , 

telephone the researcher, or send a postcard) (Einstein et al. , 1990). The use of 

naturalistic experiments raised concern about the loss of control of a number of 

varying factors (e.g., memory strategies employed by participants, compliance, 

motivation). Laboratory tasks have also been criticised as not allowing for any 

contextual integration by the participant (Einstein , McDaniel, Richardson , Guynn & 

Confer, 1995). In a laboratory task, participants may be presented with lists of 

words for learning after being instructed to perform an action whenever a particular 

target word appeared (Einstein, Holland, McDaniel & Guynn, 1992). It has been 

suggested that the findings from laboratory studies may not transfer to the real 

world , due to the high degree of sensitivity to the methodology used (Cohen , 

1996). 

There is evidence that there may be two types of prospective memory tasks. One 

of these has been described as event-based, where prospective memory is 

spontaneously cued by the occurrence of a specific external event. A second , 

time-based memory occurs as a result of a cue following the passage of time 

(Einstein et al. , 1995). Any observance of effects on prospective memory is 

thought to depend on the type of prospective task (Cherry et al., 1999). Aside from 

age-related decline in prospective memory (discussed below), ability (i.e., years of 

education), working memory span and recognition have been shown to account for 



23 

small, but significant proportions of variance in prospective memory (Cherry et al., 

1999). 

MEASURES OF MEMORY 

Neuropsychological tests allow for the impairment in distinctive components of 

memory to be identified (Lezak, 1995). One noticeable result of these attempts to 

measure the performance of separate memory functions has been the proliferation 

of testing instruments (Lezak, 1995). A further sequel to such proliferation , is the 

current lack of systematic comparisons between many of these different tests . 

This means that their interchangeability and relative usefulness is questionable 

(Lezak, 1995). This is particularly applicable in studies that to date, have 

attempted to measure the effect of HRT on memory in post-menopausal women. 

Numerous and very different neuropsychological tests have been employed 

throughout the period of study, and this variety of measures may have contributed 

to the confused and inconsistent findings (Sherwin , 1998). 

Aging , environmental influence and psychosocial variables interact in important 

ways to affect cognitive decline (Moscovitch & Wincour, 1992). With particular 

regard to the literature that has assessed the effect of HRT on memory, not all 

studies have measured or controlled for the numerous variables that may impact 

on memory performance (Sherwin , 1998). 

There has also been inconsistency in the approach to , (lleasuring memory. 

Although most studies have employed reliable and valid measures of cognitive 

functioning, not all studies have included comprehensive memory assessment 

(Sherwin, 1998). 

Furthermore, even some of the more comprehensive memory measures that have 

been utilised in the research and that are widely accepted and employed, offer 

several issues of concern. For example, the Wechsler Memory Scale-Revised 



24 

(WMS-R) and the Luria Nebraska Neuropsychological Battery Memory Scale 

(LNNB-M), have been criticised for lacking ecological validity (Makatura, Chows, 

Leahy, Castillo & Kalpakjian , 1999). 

Almost all of the studies that have investigated the possible effects of HRT in 

women have used these, or similar measures (Barrett-Connor et al. , 1993, 1999; 

Schmitt et al. , 1996; Henderson, Watt & Buckwalter, 1996). More recently, a new 

approach that utilises computerised methods that can examine the speed of 

information processing has been developed (Polo-Kantola et al. , 1998). However, 

again , the issue of ecological validity emerges. The age-related losses revealed by 

these tests may also fail to accurately reflect the degree of interference in everyday 

tasks in the real world (Rybash , Hoyer & Roodin , 1986). In the past, empirical data 

supporting a correlation between test performance on these earlier measures and 

performance in everyday functioning has been limited (Wilson , 1993). As a result , 

a fueled debate as to whether the findings that result from such testing are of any 

significance has arisen . Little or any functional representation may actually occur, 

as any compensations that individuals may make in everyday memory tasks are 

not always accounted for by these assessment measures (Rybash et al. , 1986). 

Everyday Memory has been described earlier as "memory that is evidenced in 

everyday life tasks .. . " (Tomer et al. , 1994, p.606). Makatura et al. (1999) argue 

that the psychometric instruments mentioned above tend to focus on how memory 

works rather than what an individual can do with the capacity. As a result, these 

types of instruments may fail in the assessment of areas that are often critical to 

everyday functioning . A compelling argument exists to suggest that the testing of 

memory in ways that simulate realistic everyday tasks, may provide more accurate 

information. 

Over the past few decades, there has been a rising concern regarding the lack of 

measures available for the assessment of the function of memory under realistic 

conditions. The result of such concern has seen the development of standardised 



25 

measures that present more emphasis on ecological relevance (Wilson, Cockburn 

& Baddeley, 1985; de Wall et al. , 1994). One of these, the Rivermead Behavioural 

Memory Test (RBMT) (Wilson et al., 1985), has been shown to be superior to the 

WMS-R in assessing everyday functioning when compared with behavioural 

observations (Makatura et al. , 1999). An extended version of the RBMT, the 

RBMT-E (Wilson et al., 1999) has recently been developed in order to enhance the 

original test's sensitivity in the detection of mild deficits. This extended version 

allows the opportunity to assess the everyday memory problems from a 'normal' 

and much younger sample and so will establish a starting point for future 

longitudinal studies. It is predicted to be a promising and ecologically valid 

measure of everyday memory in normal adults (de Wall et al. , 1994). 

MENOPAUSE and MEMORY - RELATED VARIABLES 

Aging 

Cognitive difficulties observed in mid aged women may be accounted for by an 

aging process that may operate independent of estrogen effect (Sherwin, 1998). 

Memory loss, particularly the encoding and retrieval of new information is a 

common concomitant of aging (Barrett-Connor et al. , 1999), although impairment in 

memory function can occur as a result of brain pathology and emotional 

disturbances as well as the aging process (Lezak, 1995). 

Longitudinal results report modest, gradual cognitive decline with aging , with an 

inflection point of around 60 years of age (Youngjohn & Crook, 1993). Age related 

deficits appear to develop in specific areas of memory (Lezak, 1995; Sherwin, 

1997). These include short-term memory (with the increased difficulty in encoding 

and retrieval of information), memory for activities, recognition, free/cued recall, 

visual/non verbal memory, and prospective memory (Smith & Earles, 1996). In 

general, older adults also demonstrate poorer performance on recall , recognition, 

memory for short paragraphs, while memory for remote past events remains intact 

(Lezak, 1995). Age differences also appear to be larger in tasks that are more 



26 

difficult and that involve more deliberate or effortful processing , while in contrast 

age differences do not appear on semantic memory tasks (Smith et al. , 1996). 

Working memory also shows evidence of an age-related decrease in speed and 

accuracy. The differences appear to be directly related to the processing efficiency 

which is facilitated by the speed with which operations can be successfully 

executed (Smith et al., 1996). In general, the measure of digit span declines only 

slightly with age, whereas word span shows a greater age decrement (Craik et al. , 

1995). However, when memorisation includes an amount longer than immediate 

storage capacity (e.g., as in story recall) , aging decrements are observable with 

larger decrements found in a delayed recall test as age increases (Lezak, 1995). 

Age differences occur in the ability to learn a new route (e.g ., non-verbal , 

visuospatial memory) , with older participants demonstrating a poorer performance. 

It does not appear that the context or the distinctiveness of materials used in the 

testing procedure impacts on the extent of the age differences observed , although 

the age effects have been found to be smaller when meaningful or familiar contexts 

are used (Hess et al. , 1996). 

Recognition Memory performance shows age-related impairments that may be due 

to a failure to encode, however, the age decrements are smaller than test of recall 

(Craik et al. , 1987; Hess et al., 1996). Cross sectional studies have demonstrated 

these decrements as a gradual decline from as early 50 years of age. The poor 

performance appears to be related to the higher rates of false positives which may 

be due to encoding strategies (Hess et al., 1996). Age differences in the 

recognition of photographed scenes are typically slight, but age related deficits are 

found when free recall or cued recall is the measure (Craik et al., 1995). 

Prospective memory studies have highlighted paradoxical age-related differences 

between naturalistic and laboratory tasks, with the latter producing an age-related 

decline, while naturalistic prospective memory tasks failed to show age differences 



27 

(Rendell et al., 1999). A likely explanation for this paradox is thought to arise from 

differences in the nature of the everyday tasks that are utilised by the different 

methods (Rendell et al. , 1999). 

Prospective memory that has been shown to decline with age, appears to occur 

especially in situations where the environment may offer few clues as reminders 

(Craik et al. , 1992). In addition , age decrements have been found on time-based 

tasks , although age differences appear to be minimal on simple prospective 

memory tasks (Einstein et al. , 1992). This is thought to be due to the degree of 

self-initiation required which may be susceptible to aging effects (Einstein et al. , 

1995). 

More recently, there also appears to be a robust age-related decline in event­

based prospective memory tasks , especially when the tasks are sufficiently 

demanding , as occurs when the number of cues are increased (West & Craik, 

1999). In the past, it has been suggested that these more complex prospective 

tasks are especially difficult for older participants, because of an increase in the 

difficulty of the retrospective memory component that exists within the prospective 

memory task (Einstein et al. , 1992). 

Failures in prospective memory have also recently been linked to an age-related 

decline in cue accessibility (i .e. , the amount of processing required to activate the 

prospective cue-action schema) . The decline in cue accessibility may reflect a 

reduced ability of adults to maintain an integrated representation of the context 

(West et al., 1999). Altogether, the extent to which age differences emerge in 

prospective memory tasks depends on the type and complexity of the prospective 

memory task and the accessibility of cues (West et al. , 1999). 

Everyday memory as a concept, has been considered to be relatively stable 

throughout adulthood until the later decades of life (Youngjohn et al., 1993). 

Furthermore, age-related decline has not been evidenced in activities that are 



28 

dependent on everyday experiences (Poon et al. , 1992). However not all studies 

have utilised a formal measure of everyday memory (Glass, 1999). 

Other possible confounding variables such as age, education , health , sleep, stress 

and mood are also likely to be involved in the possible effect of the menopause 

process on memory. It is likely that biological aging , environmental influence, 

psychosocial variables all interact in important ways to affect cognitive decline 

(Moscovitch et al. , 1992). Furthermore, chronological age has previously been 

found to be the least influential factor in accounting for total variance in memory 

performance (Arbuckle, Gold & Andres , 1986 cited in Moscovitch et al. , 1992). 

Studies have shown that medical history, educational level and lifestyle can 

contribute to significant differences in cognitive abilities among individuals within 

the same age range (Moscovitch et al. , 1992). 

However, in contrast, other studies have not provided any evidence that health 

factors or education account for age differences in memory (Salthouse, Kausler & 

Saults, 1990; Earles, Brannon & Hill , 1993). A possible explanation for the 

disparate findings with education is the changes that have occurred in the quality of 

education offered to different cohorts (Smith et al. , 1996). 

Sleep 

Sleep difficulties have been associated with menopause in that higher levels of 

difficulty are reported at later stages of the menopause transition (Slaven et al. , 

1998). Concentration and memory lapses have been associated with sleep 

deprivation , although performance nas generally been found to be better than 

anticipated (Weiten, 1995). However, increased interruptions in sleep have been 

shown to be associated with increased disruptions of mood and a reported 

decrease in general wellbeing (Peterson & Schmidt, 1999). However, it has not 

been clearly established whether some of the psychological symptoms that occur 

during menopause (e.g ., mood disruption and less satisfaction in daily tasks) are 



29 

due to lowered estrogen and progesterone levels, or are a function of disrupted 

sleep (Peterson et al. , 1999). 

Stress 

Stress is an acknowledged factor in the disruption of attention that also leads to 

impairment in a variety of functions including cognitive functioning (Weiten , 1995). 

Women enter the menopause transition at an age where it is highly likely that they 

have experienced or are currently experiencing divorce, death of spouse or parent, 

loss of employment, children leaving home and other major life events (McKinlay, 

McKinlay & Brambilla , 1987). There is now evidence from longitudinal studies that 

appears to add some weight to the hypothesis that these changes in life events 

may be associated with the observable psychological distresses noted during 

menopause (Slaven et al ., 1998). 

Mood 

Emotional !ability, depression and a decreased sense of well-being are common 

complaints during the menopausal transition (Mayeaux et al ., 1996). Controversy 

exists as to whether these observed changes in mood constitute a menopause­

related mood disorder, and there does not appear to be clear evidence of any 

increase in depression that is related to the menopausal years (Ballinger, 1990). 

However, there is the suggestion that women with known histories of depressive 

symptoms may be at increased risk during menopause which may indicate an 

increased risk for this group of women (Pearlstein , 1995). 

Alteration in mood has been associated with the reduction of estrogen that occurs 

at menopause. Two neurotransmitters (serotonin and norepinephrine) are involved 

in the regulation of mood , and biochemical evidence now suggests that estrogen 

effects may influence this regulation of mood (Halbreich, 1997). 



30 

The idea that a reduction in estrogen may predispose post-menopausal women to 

depression remains controversial and has not always been supported by 

epidemiological studies (Halbreich, 1997). In addition , empirical data has not 

always supported estrogen as an effective treatment for major depression , 

although interestingly, it has been reported to have a positive effect on healthy, non 

depressed , post-menopausal women (Ditkoff et al. , 1991 ). 

However, HRT does appear to play a role in the amelioration of some depressive 

symptoms. Chatel , Fugere, Bossinette & Berube (1996) found depression scores 

improved, although not significantly, in a group of 57 women attending a 

menopause clinic. Also , Cagnacci , et al. (1997) found that estrogen use improved 

depression and anxiety as measured by the Self -Evaluation Depression Scale, 

although a later study failed to demonstrate any beneficial effect of HRT on 

depressed mood (Cagnacci , Neri , Tarabusi , Volpe & Facchinetti , 1999). 

Between these two reported studies, and contributing to a continuing controversy, 

a meta-analysis of the effect of HRT on depressed mood (Zweifel & O'Brien , 1997) 

concluded that HRT was effective in reducing depressed mood in menopausal 

women . In this meta-analysis, methodology, such as controlled design , sample 

sizes and valid measures of depression were all reported as adequate. A later, 

more limited review (Archer, 1999) further supports this stance, concluding that 

there is suggestive evidence that estrogen may be an effective treatment for 

depression in both perimenopausal and menopausal women . 

It is possible that the variable definitions of depression and menopause contribute 

to the disparity of the findings, making interpretation problematic. In addition , 

sample characteristics, such as women presenting at menopausal clinics being a 

self-selected population and the likelihood of these women having experienced 

previous depressive episodes, may further compound interpretation difficulties 

(Pearlstein, 1995). Thus, research samples may not always be representative of 

the menopausal population of women. However, on the whole, there is increasing 



31 

empirical support for a positive effect of HRT on mild to moderate depressive 

symptoms that may accompany the menopause transition . 

A link between depression and cognitive abilities such as memory has been 

acknowledged , although significant memory impairments have not always been 

associated with depression , and research is often contradictory (Lezak, 1995). For 

example, no differences have been found between depressed unmedicated 

women and matched controls (Gas & Russell , 1986). Also , although non 

depressed participants have demonstrated a trend towards higher scores, 

significant differences in tests of verbal short term retention or learning have not 

been found (Marsh, Marsh & Johnson , 1987, cited in Lezak, 1995). In addition , 

depressed patients have been shown to perform as well as non-depressed patients 

in tests of speed , recognition memory and abstraction (Cole & Zarit, 1984). 

However, depressed patients have been shown to demonstrate cognitive 

impairment and forgetfulness. Impaired performance on attention and immediate 

memory tasks has also been shown, with the severity of impairment related to the 

depression (Cohen , Weingarter, Smallberg , Pickar & Murphy, 1982). Facial 

recognition has been found to be significantly impaired relative to test norms, 

although verbal learning and memory measures remained at normative values 

(Sweeney, Wetzler, Stokes & Kocsis , 1989). Difficulties in concentrating , and 

changes in cognition such as problems in thinking and making decisions, are listed 

as symptoms of depression in the DSM IV (Kaplan & Sadock, 1998). 

It is thought that the memory dysfunction demonstrated in association with , or as a 

result of depression is more likely to be a reflection of poor encoding than a 

learning problem. Deficits demonstrated in both verbal and visuospatial material 

have been attributed to retrieval on free recall , rather than learning (Massman, 

Delius, Butters, Dupont & Gillian, 1992). 

The above review highlights the complexity of the relationship between mood (in 

particular depression) and cognitive ability. The association of menopause with 



32 

mood !ability is well documented, and clinically observed, but remains controversial 

(Halbriech , 1997; Mayeaux et al ., 1996). Much of the literature, but not all , 

supports the presence of significant and specific impairments in cognitive ability in 

clinically depressed individuals (Lezak, 1995). HRT use has been shown to 

improve mild to moderate depressive symptoms (Cagnacci et al ., 1997; Archer, 

1999). In view of the evidence of such an association between mood and cognitive 

ability, research that investigates the effects of other variables on memory must 

consider and control for the possible confounding effects of mood . 

SUMMARY 

The association between the construct of memory and the processes of aging and 

menopause is a complex one. The period of menopause has been linked to 

memory deficits, with women undergoing surgical menopause in particular, 

showing a negative affect on short-term verbal memory. Generally, the 

measurement of memory has often lacked both external and ecological validity 

(i .e., the degree to which tasks used in research are similar to the tasks of 

everyday life) (Hess et al ., 1996). The presence of a number of confounding 

variables has increased the difficulty in assigning clear and independent 

associations of the effect of these variables in previous studies. In particular, the 

accurate assessment of affect or mood state is important when assessing the 

memory of mid-life women. Somatic symptoms, such as fatigue , insomnia and 

changes in weight, that occur during the menopausal transition are frequently used 

to index depressive or anxious feelings. Thus far, there has been a surprising lack 

of research investigating the possible effect of HRT on older women's cognitive 

functioning , particularly in regard to the performance of everyday tasks. Such 

tasks include the ability to remember names, to recognise faces , to recall where 

personal belongings have been left, to attend to and remember conversations, and 

to remember to remem~er. The ability to perform these tasks are all important 

aspects of everyday memory functioning. Ultimately, the preservation of these 

abilities determine the quality, and independence of living maintained by women as 



33 

they continue through, and past the menopause transition. This is of increasing 

relevance as society prepares for an increased aging population and the projected 

associated increase in dementia, especially of Alzheimer's type. Investigation of 

the possible relationship between HRT use and everyday memory tasks is one of 

the aims of the present study. 



34 

CHAPTER4 

PURPOSE AND FORMULATION 

This present study was undertaken to assess the relationship between HRT use 

and mid-aged women's cognitive status. The study was planned in order to 

establish the suitability and cultural sensitivity of measures that to date, had 

primarily been used with overseas populations. In particular, the study would pilot 

the use of The Rivermead Behavioural Memory-Extended Test (a newly created 

measure of everyday memory tasks) on a New Zealand sample of mid-aged 

women . The data would also provide some estimates of the relationships between 

mood, HRT use and memory in a sample of mid-aged New Zealand women . 

AIMS OF THE PRESENT STUDY 

• The first aim of the study is to investigate the effect of HRT use on the everyday 

memory ability of New Zealand women between the ages of 40 and 60 years. 

More specifically, to test whether there is a significant difference in everyday 

memory performance between mid-aged women using HRT and women not 

using HRT, while taking into account the effects of age , mood , education , 

general health and stress. 

• A second aim of the study is to evaluate the use of a measure of everyday 

memory (The Rivermead Behavioural Memory Test-Extended Version) on a 

sample of mid-aged women. 

HYPOTHESES 

• The first hypothesis is that women who use HRT will show better everyday 

memory ability than those women who do not use HRT, while controlling for 



35 

age, affect, mood, health, education and stress. In addition , it is expected that 

verbal memory will show the greatest difference between the two groups. 

• The second hypothesis predicts that memory ability will decline with increasing 

age. 

• The third hypothesis predicts that age will moderate the relationship between 

HRT use and memory ability. More specifically, it is predicted that differences 

in memory ability between HRT users and non HRT users will be greatest for 

older women. 

• The fourth hypothesis predicts that mood will be positively related memory 

ability. More specifically, positive mood will be associated with better memory 

and conversely, negative mood will be associated with poorer memory ability. 

• The fifth hypothesis predicts that mood will moderate the relat ionship between 

HRT use and memory ability. 

DESIGN 

The study employed a cross-sectional , correlational , between subjects design , 

comparing a sample of New Zealand women aged between the ages of 40 and 60 

years who were either HRT users or non HRT users. 



36 

CHAPTER 5 

METHOD 

PARTICIPANTS 

A total of 104 women with an age range of 40-60 years (mean age, 51 .67 years) 

who volunteered to participate in the project were studied . Of these women , 53 

(51 %) were HRT users and 51 (49%) were non-users. Demographic data for the 

total sample are summarised in Table 1. To avoid duplication and for ease of 

comparison, data for the two HRT use groups are included . Twenty-six percent 

considered themselves pre menopausal , 39.4% as menopausal , and 33 .7% 

identified themselves as post-menopausal. All the women were functioning 

independently in the community and 76.9 % were employed in a full or part -time 

work. Seventy per cent reported participating in regular exercise, while 63 .5% of 

participants were taking some form of prescribed or 'over the counter' medication . 

Fifty-five percent of the sample had received an excess of 12 years education , and 

of this group, 60% were being treated with HRT. 

Participants were recruited by a number of methods including speaking to health 

professionals (Gerontologists) at a local hospital in Christchurch , and by 

advertisements (see Appendix 1) placed in local newspapers in Christchurch and 

Palmerston North . The following inclusion criteria were used: 

1. Women aged between 40 and 60 years of age 

2. All ethnic groups of New Zealand citizens. 



37 

Table 1. 
DemograQhic & Health Characteristics of HRT use GrouRS and Total SamRle. 

HRT use non HRT use Total 
(n=53) (n=51) (n=104) 

Age 52 .34 years 50.98 years 51 .67 years 
(Mean) (Mean) (Mean) 

Ethnicity 
NZ European 48 (51 .6%) 46 (48.4%) 90.4% 
Other Ethnicity 5 (9.4%) 5 (9.8%) 9.6 % 

Education 
High School 20 (37.7%) 26 (51 .1%) 44.2% 
(<11yrs) 
Diploma!Tertiary 33 (62 .3%) 25 (49.0%) 55.8% 
(> 12yrs) 

EmQloyment 
full/part-time paid 43 (81 .1%) 37 (72.5%) 76.9% 
full-time unpaid 10 (18 .9%) 14 (25.4%) 23 .1% 

Exercise 36 (67.9%) 36 (70.6%) 69 .2% 
Nil exercise 17 (32 .1%) 15 (29.4%) 30.8% 

History: 
Head Injury (yes) 10 (18.9%) 9 (17.6%) 18.3% 

(no) 43 (81 .1%) 42 82.4%) 81 .7% 

Smoking 
Never 27 (50.9%) 36 (70.6%) 60.6% 
Past 21 (39.6%) 12 (23 .5%) 31 .7% 
Present 5 (9.4%) 3 (5 .9%) 7.7% 

Family neurological 
Strokes 14 (26.4%) 12 (23.5%) 25.0% 
Alzheimer's dementia 5 (9.4%) 6 (11 .8%) 10.6% 
Other dementia 3 (5 .6%) 2 (3.7%) 4.8% 
Neu rolog ica I 31 (58.5%) 32 (62.7%) 60.6% 



38 

An IQ mean (122, SD. 6.35) was established for the total sample from the scores 

on the NART, a measure of crystallised intelligence. The NART scores yielded a 

mean of 6.51 (SD. 5.11) errors for the total sample. Ninety-six percent of the entire 

sample were classified as above average (predicted IQ 111 and above), and 3.2% 

as average group were classified as average (predicted IQ 90-110) . The high 

percentage that is classified as above average may not be representative of the 

population of mid-aged women. Table 2 shows the means and standard deviations 

of the NART and IQ for the total sample. 

Table 2. 
NART Errors and established IQ classification for Sample. 

NART errors 

IQ 

6.51 

122 

Mean (SD) 
(n=104) 

(5 .11) 

(6.35) 

The histories of personal illness for the sample are shown in Table 3. The 

numbers of women who used concomitant medication are shown in Table 4. Both 

tables are expressed as a percentage of either the HRT or non HRT group as well 

as for the total sample. 

Table 3. 
Personal Medical History of HRT use Groups and Total Sample. 

Personal Illness 
HRT users non HRT users Total 

(n=53) (n=51) (n=104) 

Diabetes 2 (3.7%) 2 (1 .9%) 

High BP 11 (20.7%) 12 (23.5%) 23 (22.11%) 

Heart Disease 1 (1.9%) 2 (3.9%) 3 (2.88%) 

Epilepsy 1 (2%) 1 (0.96%) 

Cancer 4 (7.5%) 5 (9.8%) 9 (8.65%) 

Depression (history) 16 (30.2%) 5 (9.8%) 21 (20.19%) 

PMT/symptoms 21 (39.6%) 22 (43.1 %) 43 (41.34%) 



39 

Table 4. 
Concomitant Medications Use bY'. HRT use GrouQS and Total SamQle. 

HRT use non HRT use Total 
(n=53) (n=51) (n=104) 

Asthma 4 (7%) 2 (3 .9%) 6 (5 .76%) 

Anti-inflammatory 5 (1%) 4 (7 .8%) 9 (8.65%) 

Memory enhancing 6 (11.7%) 6 (5.76%) 

Antidepressants 4 (7%) 3 (5.8%) 7 (6.73%) 

Vitamins/Minerals 12 (22.6%) 13 (25.5%) 25 (24 .03% 

Calcium 10 (18.8%) 5 (9 .8%) 15 (14.42%) 

BP medication 11 (20.7%) 8 (15.6%) 19 (18.26%) 

Thyroid 4 (7 .5%) 4 (7 .8%) 8 (7.69%) 

Peptic ulcer 2 (3.7%) 1 (1.9%) 3 (2 .88%) 

Sleep 2 (3 .7%) 1 (1.9%) 3 (2 .88%) 

Pain 1 (1.8%) 1 (0.96%) 

St. John Wort 1 (1.9%) 1 (0 .96%) 

MEASURES 

The following scales (see Appendix 2) , were chosen as a set of measures for the 

study. As well as the two psychological constructs of everyday memory and mood , 

scales were chosen to control for the effect of known possible confounding 

variables. Each will be described in turn . 

1. Measuring Everyday Memory 

The Rivermead Behavioural Memory Test - Extended Version (RBMT-E) . (Wilson , 

Clare, Cockburn, Baddeley, Tate & Watson , 1999). 

The RBMT-E has been described as a standardised and reliable measure of 

everyday memory (Wilson et al., 1999). To date, it has not been used with a New 

Zealand population. The original Rivermead Behavioural Memory Test (RBMT) 

(Wilson, Cockburn & Baddeley, 1985) was designed to predict everyday memory 



40 

problems in people with acquired , non-progressive brain injury, and also to monitor 

change over time (Wilson et al. , 1999). The RBMT has been shown to be a useful 

tool in the measurement and prediction of everyday memory problems (van Balen , 

Westzaan & Mulder, 1996). 

Inter-rater reliability has been shown to be 100% for both profile and screening 

scores. Also, parallel-form reliability has been demonstrated between versions A 

and B, C and D as 0.86, 0.83 and 0.88 respectively (Wilson et al. , 1985). The 

RBMT has been used with New Zealand populations, has excellent face validity 

and a recognised ecological validity (Fraser, Glass & Leathern , 1999). Normative 

data for a New Zealand population of 131 elderly people (60-89) recently 

established (Fraser et al. , 1996). In addition to issues of reliability and validity, the 

design of four different versions allowed repeated assessments while minimising 

practice effects (Wilson et al. , 1989). 

The extended version of the Rivermead (RBMT-E) has been developed in order to 

enhance the sensitivity of the RBMT by increasing the difficulty of the test. 

Versions A and B of the RBMT have been combined to make Version 1 of the 

RBMT-E, and Versions C and D have been combined to make Version 2 of the 

RBMT-E. This has resulted in an extension the original RBMT from that of a 

screening test, to a test providing a sensitive measure of memory within a normal 

range (de Wall et al. , 1994). 

The extended version has been designed to follow the original structure so as to 

capitalise on the established validity and sensitivity of the original RBMT (Wilson et 

al., 1999). In a pilot study (de Wall et al., 1994) found that the RBMT-E 

significantly discriminated between a middle-aged (40-55 years) and older group 

(65-79 years), even when these differences were small as measured by the 

Warrington Recognition Memory Test (Lezak, 1995). Participants also included 18 

individuals from African-Caribbean and Asian origin to ensure cultural 

appropriateness and sensitivity (de Wallet al., 1994). 



41 

The RBMT-E (Wilson et al. , 1999) is comprised of two parallel versions that have 

been shown to be sensitive to age and IQ effects in non-brain participants. 

Parallel-form reliability was investigated by administering the 2 versions of the test 

to 191 control subjects aged between 16-76 years. Overall mean scores were very 

similar between the two versions (Wilson et al. , 1999), although tests of 

significance were not reported in the manual. 

The test has been shown to be capable of detecting more subtle memory deficits 

than the original RBMT. In a comparison of 45 neurologically impaired people on 

both the original and the RBMT-E , the newly extended version separated the 

RBMT scores into good, average, poor and impaired subgroups (Wilson et al. , 

1999). 

In discussing the assumption that the RBMT-E is a valid measure of everyday 

memory, de Wall et al. (1994) cite the ecological validity of the original RBMT. This 

was firmly established with the use of contrasting groups for whom everyday 

memory problems were or were not prominent, and then subsequently validated 

against hours of careful observation . Correlations between therapist observations 

and scores obtained on the RBMT were significant (=0.75). 

Materials: 

The RBMT-E consists of a manual with instructions for administration and scoring 

of the test. A test materials book contains the sub-tests and allows the tester to 

read the instructions while the participant views the stimuli. In addition , 

supplementary items include a large picture card with 15 items (picture 

recognition) , a message envelope (route & messages) and a timer. A scoring 

sheet (Appendix 2) , which is also a procedural guide is included. 

Sub-tests: 

The RBMT-E is comprised of 11 sub-tests: 



42 

* Sub-test 1 & 2 - First and Second Names. 

The participant is shown three photographic portraits and asked to remember the 

first and second names of all three people in the photographs. The participant is 

required to recall the names at the end of the test. Each name recalled without a 

prompt is scored 2, recalled after a prompt is scored 1. 

* Sub-test 3 - Belongings. 

Two possessions belonging to the participant are borrowed and secreted in a desk 

drawer and a cupboard (or suitable alternatives). The participant is requested to 

ask for the belongings at the end of the test session, and to remember where they 

have been hidden. Each item recalled without a prompt is scored 2, recalled after 

a prompt is scored 1. 

* Sub-test 4 - Appointments. 

This sub-test assesses response to cueing . The tester sets an alarm for 20 

minutes time. The participant is required to ask two questions relating to the near 

future , when the alarm sounds. For Version 1, the questions are "When do I have 

to see you again?" and "When does this session end?" For Version 2, the 

questions are "When will I know the results of the test?" and "What time do we 

finish today?" The participant is told to remember to ask both questions when the 

alarm sounds. If the participant does not ask the questions spontaneously when 

the alarm sounds, prompts are given (e.g. , "What were you going to do when the 

alarm rings?") . Each question asked spontaneously scores 2, each question after 

a prompt is scored 1. 

*Sub-test 5 - Picture Recognition (Presentation) . 

The participant is shown 20 line drawings simultaneously on a large sheet of card . 

Participants are requested to memorise as many as they can in 15 seconds, and to 

recognise these later on from a set of 40. The tester informs the participant of the 

time after 5 and 10 seconds. At 15 seconds, the tester informs the participant that 



43 

'time's up'. The raw score is obtained by deducting the number of false positives 

from the number of original pictures correctly identified. Maximum score=20 

*Sub-test 6 - Story (Immediate) . 

The participant is asked to listen to a short passage of prose being read out, and 

then to immediately recall as much of it as possible. (The participant is asked to 

recall the story again later in the testing procedure, without a second reading) . The 

authors (Wilson et al ., 1999) make particular note that certain names and possibly 

a few words are culture specific. The tester is advised to change names/words to 

suit the culture of the participant. Later, after appointments, the tester asks the 

participant to recall the prose passage (story delayed). For scoring , the story is 

divided into 21 ideas with each idea that is recalled perfect or close =1 . Guidelines 

are given in an appendix for scoring 1/2 a point for partial or approximate 

synonyms. Maximum score 21 . 

* Sub-test 7 - Face Recognition (Presentation) . 

The participant is shown pictures of 15 faces , one at a time for 3 seconds each . 

The participant is asked to decide whether the person in the photo is under or over 

40 years old . The participant is later required to select the original 15 from a set of 

30. Scoring is obtained by deducting the number of false positives from the number 

of original faces correctly identified. Maximum score 15. 

* Sub-test 8 - Route (Immediate) and Sub-test 9 - Messages (Immediate) . 

The tester traces a short route within the room . The route comprises of 7 sections. 

The participant is asked to retrace the route immediately. An example is given in 

the manual - two chairs, a door, a window, a heater, table and noticeboard. After 

the story delayed , the participant is required to retrace the original route with the 

messages ('Route' and 'Messages Delayed'). If the message envelope and book 

are not spontaneously picked up, a prompt is again given (e.g ., "I took two things 

with me"). Maximum score 15. 



44 

*Sub-test 10 & 11 - Orientation and Date. 

The participant is asked 13 questions pertaining to their present situation (e.g., 

"what year, month, day, time, date; name of place, city, age and year of birth ; 

present and previous Prime Ministers, present and previous Presidents of the 

USA?") . Maximum score= 14, with 2 points awarded if date is correct. 

After completion of the story, route and messages (delayed) , the original three 

portrait photographs are represented one at a time. Initial letters are provided as 

prompts if response is not spontaneous. The test ends with the tester informing 

the participant, "We have finished this test" as a cue to see of they remember to 

ask for their belongings, and where they has been placed . If the participant does 

not request their belongings, the tester prompts with "You were going to remind me 

to give you some things" . 

Scoring : 

Because sub-test raw scores have different minimum and maximum scores , 

conversion to sub-test profile scores is required . The development study (Wilson 

et al. , 1999) used box plot analysis for this purpose and further adjustments were 

made for observed age and IQ effects in this study. In the present study, 

adjustments were performed as per the manual. Sub-tests that were adjusted for 

age effects were ; 'Route Immediate' and 'Route Delayed '; sub-tests adjusted for IQ 

effects were; 'First Names', 'Second Names', 'Story Immediate' and 'Story 

Delayed'. The raw scores of sub-tests 3 and 4 (i .e., 'Belongings and 

Appointments') were summed to create a 'Prospective Memory' sub-test before 

conversion to a profile score. Individual Profile sub-tests have a maximum score of 

4 points and are summed to create a total profile score (maximum - 48 points) . In 

addition , five classifications of an overall profile of memory are calculated. These 

classifications are; impaired (0-18) , poor (19-27) , average (28-36) , good (37-42) 

and exceptionally good (43-48)(Wilson et al. , 1999). 



45 

2. Measuring Mood 

Mood was operationally defined first as a mild , usually transitory emotion (Chaplin , 

1985), and this was measured by the Profile of Moods Questionnaire (McNair, Lorr 

& Droppleman, 1992). Second, the construct of mood was extended to embrace 

the two dimensions (positive and negative) of the broader structure:- affect. This 

was measured by the Positive and Negative Affect Scale (Watson , Clark & 

Tellegen , 1988). 

Profile of Moods Questionnaire (POMS). 

The POMS has been used widely as a measure of transient mood states 

(Nyenhuis , Yamamoto , Luchetts , Terrien & Parmenter, 1999) and also for 

depression (Lezak, 1995). 

The POMS utilises a specific affects approach which emphasises the multiple 

unique dimensions of mood and includes constructs such as Depression , Tension , 

Anger, Vigor, Fatigue, and Confusion . Examples of descriptive feelings are 

Depression , 'unhappy', 'sad ', and 'blue; Tension , 'shaky', and 'panicky'; Anger, 

'peeved ' and 'annoyed '; Vigor, 'lively' and 'active; Fatigue, 'worn out', and 'listless'; 

Confusion , 'muddled ' and 'forgetful '. Participants are asked to rate on a 5 point 

scale (0 , 'Not at all ', to 4, 'Extremely') (see Appendix 2). The summation of the 

sub-scale measures (vigor is weighted negatively), allows a description of a total 

mood disturbance, with a high score indicating a greater mood disturbance (McNair 

et al. , 1992). 

The POMS has shown good reliability and validity, and factor analyses have 

confirmed consistency with reasonably good test-retest correlations (Lezak, 1995). 

Convergent and discriminant validity has been shown to be good (Nyenhuis et al., 

1999), with the POMS scales more highly related to corresponding measures of 

mood than non-mood scales. Furthermore, the POMS correlated highly with the 

BDI depression scale, and the visual analogue mood scale. 



46 

The POMS has been shown to be a reliable and valid measure for mood states in 

older adults (Gibson, 1997), and has been used in several studies involving 

menopausal women (Kimura et al. , 1995; Slaven & Lee, 1994, 1997, 1998; 

Schiffman, Satteley-Miller, Suggs & Graham 1995; Adams, Cartwright, Ostrove, 

Stewart & Wink, 1998). 

Positive and Negative Affect Scale (PANAS). 

The PANAS (Watson , Clark & Tellegen , 1988) is a 20-item scale, developed to 

measure affect. In particular, the PANAS was designed to capture the two 

identified , highly distinctive dimensions (positive affect, and negative affect) . Such 

distinction has been supported by low correlations between the two dimensions 

(range: -0 .12 to -0.23) (Watson et al. , 1988). 

The PANAS lists 20 mood descriptions (10 describing a positive affect scale, 

'Positive Affect') and (10 describing a negative affect scale, 'Negative Affect') . 

Participants rate on a 5 point scale (very slightly or not at all , a little, moderately, 

quite a bit and extremely) , the extent to which they have been experiencing each 

(see Appendix 2) . Examples of 'Negative Affect' include 'distressed ', 'upset' and 

'guilty', while 'Positive Affect ' examples include 'excited ', 'strong ' and 'enthusiastic', 

with a high score indicating a high degree of positive affect. The positive and 

negative affect items are summed separately for scoring (range 10-50 for each 

scale) , to create a 'Positive' and 'Negative' Affect variable label in the present 

study. High scores mean a greater degree of negativity or positivity. 

Various time instructions can be used with the PANAS ranging from the present 

moment to feelings generally. For the present study, participants were asked to 

indicate the extent of each feeling 'during the past week'. The PANAS also 

showed good convergent validity with a variety of other brief affect measures (0.76 

- 0.92) (Watson et al., 1988). 



47 

The normative data for the PANAS was gathered in the main from college student 

population (Watson et al., 1988). In an effort to report generalisability to a non­

student population , 164 non-student adults were also tested . Alpha reliabilities of 

the 'Positive' and 'Negative Affect' were 0.86 and 0.87 respectively, with a 

correlation between the scales of .09 and Watson et al. (1988) believe the PANAS 

is able to offer useful information in adult samples. Mackinnon et al. (1999) , 

examined the applicability of the PANAS across the life span, with a large sample 

(n=2651) 18-79 year olds. Factor structure and factor correlations did not appear 

to change with age. 

3. Measuring Confounding Variables 

Copies of the measures chosen to control for the effect of possible confounding 

variables are shown in Appendix 2. These included the Women's Health 

Questionnaire (WHQ) and the Short-Form Health Survey (SF 36). The National 

Adult Reading Test (Nelson , 1982; Nelson & Williston , 1991) was used in order to 

provide a pre-morbid estimation of intelligence, and the Digit Span (Wechsler, 

1981) as a measure of attention and short-term memory. A measure of the 

number of life change events, the Social Readjustment Rating Scale (Holmes & 

Rahe, 1967) was employed to control for the possibility of stress. 

Women's Health Questionnaire (WHO). 

The Women's Health Questionnaire was specifically developed by Hunter (1992) to 

measure the wide range of physical and emotional symptoms experienced by mid­

life women aged 45-65 years. It was designed to identify changes that may occur 

throughout the menopause transition , and to allow the assessment of the individual 

effects of symptoms that may arise (Hunter, 1992). 

The WHO consists of 36-items that reflect somatic, vasomotor, memory, 

concentration, sexual behaviour, sleep, menstrual, depressed, and anxious mood 

symptoms which the participant rates on a 4 (0-3) point scale (No, Not at all; No, 

not much; Yes, sometimes, and Yes, definitely). (see Appendix 2) . Examples of 



48 

these sub-scales are; somatic, 'My stomach feels bloated'; vasomotor, 'I suffer 

from night sweats'; memory and concentration, 'my memory is poor' , I have 

difficulty in concentrating'; sexual behaviour, 'I have lost interest in sexual activity' ; 

sleep, 'I have difficulty in getting off to sleep', menstrual, 'I have heavy periods'. In 

the present study, the depressed and anxious factors were omitted as they were 

thought to be adequately assessed with the POMS and the PANAS. Hunter (1992) 

acknowledges that not all the scales will be of interest in every study, and that 

smaller sub-scales may be omitted . Furthermore, the presence of a separate 

factor to assess sleep difficulties is a particular strength of the WHO when 

considering the performance of memory in mid-aged women. For scoring, 

symptom items for each sub-scales were summed and divided by the number of 

items in each sub-scale. A total score allows an overall measure of well-being with 

a higher score reflecting more disturbance. 

Several studies involving menopausal women have utilised the WHO (Wilkund et 

al., 1992; Greensmith , 1991 ; Slaven et al. , 1997, 1998), and it has been found to 

be a sensitive measure of response to hormone replacement therapy. Test - retest 

reliability was assessed after a two-week period and correlations between sub­

scales ranged between 0.69 and 0.96 . 

Short-Form Health Survey Questionnaire (SF-36). 

The SF-36 (Ware & Sherbourne, 1992), was constructed to survey health status 

and designed for use in clinical practice or research, health policy evaluations and 

general population surveys. The SF-36 is a 36-item questionnaire that measures 

eight multi-item dimensions tapping functional status, well-being and overall 

evaluation of health. Functional status evaluates physical and social functioning, 

as well as role limitations occurring as a result of physical and/or emotional 

problems. Physical functioning lists 10 activities, and asks the participant to rate 

how their health limits them on a 3 point scale (Yes, limited a lot; Yes, limited a 

little, and No, not limited at all). Social functioning asks the participants to rate on a 

5 point scale (from all to none), how much time physical health and emotional 



49 

problems interfered with their social activities. An emotional and physical role 

limitation scale allow a yes/no rating (range 1-2). The well-being scale assesses 

general mental health (e.g., 'Have you been a very nervous person ') and bodily 

pain (very severe to no bodily pain) and vitality (e.g ., Do you feel full of life). A 

further scale evaluates general health perception , and also includes one non­

scaled item that asks participants about health change over the past year 

(Jenkinson , Wright & Coulter, 1994). High scores on 'Functional Status ', 

'Wellbeing ' and 'Health Evaluation ' indicate the best possible health state. 

The questionnaire is designed, and has been used successfully for self­

administration , telephone administration , or administration during a personal 

interview (Ware et al. , 1992). McHorney, Ware, Lu , & Sherbourne (1994) 

undertook tests of data quality, scaling assumptions and reliability across diverse 

patient groups, and found that all scales passed tests for item-internal consistency 

and item-discriminant validity. Reliability coefficients ranged from 0.65 to 0.94 

across scales. Jenkinson et al. (1994) found internal consistency to be good , with 

alpha values > 0.8 for all dimensions of the SF-36 except for the social functioning 

scale. Jenkinson et al. (1994) also reported the ability of the SF-36 to discriminate 

between different groups of respondents (those reporting poor to excellent health) . 

As the authors assert, this suggests that the questionnaire is a suitable instrument 

for use in homogeneous groups. 

The validity of the SF-36 has been criticised on the grounds that it does not contain 

items about sleep (Hunt & McKenna, 1993). To investigate this further, Lyons, 

Fielder & Littlepage (1993) used the SF 36 in a study on health status, and 

additionally asked questions of participants regarding any 'problems sleeping' 

experienced . Participants that reported treatment for a sleeping disorder were 

found to have lower scores on all the SF 36 variables and Lyons et al. (1993) 

suggest that the lack of questions about sleep does not detract from the validity. A 

later study (Zammit, Weiner, Damato, Sillup, & McMillan, 1999) confirmed the 

findings of Lyons et al. (1993) . Participants with insomnia also obtained lower 

mean scores on sub-scores of the SF-36 indicating widespread impairment. 



50 

Although the validity of the SF-36 is confirmed , without specific sleep questions, 

the association of sleep difficulties with memory performance is difficulty to identify. 

Sleep difficulties are a common complaint of menopausal women and are also 

correlated with poor cognitive performance (Hunter, 1992; Weiten , 1995). For the 

purposes of the current study, sleep difficulties were assessed as a separate factor 

within the WHO as outlined above. 

The acceptability, reliability and validity of the SF-36 has been assessed in a N