reduced verbal memory retention is unrelated to sleep disturbance during pregnancy

Reduced Verbal Memory Retention is Unrelated to SleepDisturbance During Pregnancyap_76 196..208

Danielle L Wilson,1,2 Maree Barnes,2,3 Lenore Ellett,4 Michael Permezel,4,5 Martin Jackson,1 and Simon F Crowe1

1School of Psychological Science, La Trobe University, 2Institute for Breathing and Sleep, Austin Health, 3Department of Medicine, University of

Melbourne, 4Department of Obstetrics and Gynaecology, Mercy Hospital for Women, and 5Department of Obstetrics and Gynaecology, University of

Melbourne

This study investigated episodic and procedural memory retention in early and late pregnancy and whether memory retention was related tosleep disruption. Twenty-six women in the third trimester of pregnancy, 20 women in the first trimester of pregnancy, and 24 non-pregnantcontrols were administered a battery of verbal and visual episodic memory tasks and two procedural memory tasks before undergoing anovernight sleep study. Memory retention was assessed the following morning. Results indicated that as compared with controls, both pregnantgroups had reduced retention in verbal episodic memory but were unimpaired on visual and procedural memory tasks. The pregnant womenalso demonstrated significant disruption of sleep patterns. Reduced verbal memory retention during pregnancy was not attributable to anymeasure of sleep; however, small correlations between some indices of sleep and memory do not allow full dismissal of the sleep-dependentmemory consolidation hypothesis.

Key words: attention; consolidation; episodic; pregnant; procedural; progesterone.

Memory complaints are commonly reported during pregnancy(Janes, Casey, Huntsdale, & Angus, 1999; Parsons & Redman,1991), with deficits demonstrated on word list learning(Buckwalter et al., 1999; de Groot, Vuurman, Hornstra, & Jolles,2006; Sharp, Brindle, Brown, & Turner, 1993), paragraph recall(Keenan, Yaldoo, Stress, Fuerst, & Ginsburg, 1998) semanticfluency (de Groot, Hornstra, Roozendaal, & Jolles, 2003), andpriming (Brindle, Brown, Brown, Griffith, & Turner, 1991;Sharp et al., 1993). However, this effect is not consistently

observed (Casey, Huntsdale, Angus, & Janes, 1999; Christensen,Poyser, Pollitt, & Cubis, 1999). One common explanation formemory deficits during pregnancy relates to hormonal changes,but no consistent associations between memory impairmentand pregnancy hormones have been noted (Buckwalteret al., 1999; Silber, Almkvist, Larsson, & Uvnas-Moberg, 1990).The cause of memory problems during pregnancy is yet to bedetermined.

Another frequently made complaint during pregnancy is thatof sleep disruption. While correlations between crude measuresof self-reported sleep disturbance and verbal memory have beennoted (Christensen, Leach, & Mackinnon, 2010; Keenan et al.,1998), as yet no study has focused on objective measuresof sleep as contributing factors to memory difficulties duringpregnancy.

Polysomnography (PSG) studies undertaken during preg-nancy have consistently shown that sleep efficiency (time spentsleeping as a percentage of time spent in bed) is reduced in thepregnant state and deteriorates as pregnancy advances (Brunneret al., 1994; K. A. Lee, Zaffke, & McEnany, 2000), mostly due toincreased time awake after sleep onset (Brunner et al., 1994;

Correspondence: Simon F. Crowe, School of Psychological Science, LaTrobe University, Bundoora, Vic. 3086, Australia. Fax: +61 3 9479 1956;email: [email protected]

This research was funded by La Trobe University and the Austin HealthMedical Research Foundation. The authors have indicated no financialconflicts of interest.

Accepted for publication 10 April 2012

doi:10.1111/j.1742-9544.2012.00076.x

What is already known on this topic

1 Pregnant women commonly report memory difficulties, butwhether real memory deficits exist is controversial.

2 Sleep disturbance is a frequent complaint made during preg-nancy, and objective measurements have mostly supportedsuch reports.

3 Theories that propose that memory consolidation occurs duringsleep have received considerable empirical support, but thisproposal remains contentious.

What this paper adds

1 Pregnant women have reduced verbal episodic memory reten-tion, but visual and procedural memory is unaffected.

2 Objective measures of sleep during pregnancy reveal reducedtotal sleep time, more night awakenings, and less deep sleep infavour of more light non-restorative sleep.

3 The combination of reduced verbal memory and sleep distur-bances during pregnancy was not consistent with theoriesof memory consolidation during sleep, and reduced verbalmemory could not be explained by other factors such as atten-tion deficits, mood, or hormone level.

bs_bs_banner

Australian Psychologist 48 (2013) 196–208© 2012 The Australian Psychological Society

196

Driver & Shapiro, 1992; Hertz et al., 1992). The alterations inrapid eye movement (REM) sleep associated with pregnancyhave ranged from a reduction in comparison to non-pregnantcontrols (Driver & Shapiro, 1992; Hertz et al., 1992), a reductionfrom early to late pregnancy (Brunner et al., 1994), throughshowing no significant differences (K. A. Lee et al., 2000; Schorret al., 1998). Slow wave sleep (SWS) is often decreased com-pared with prepregnancy baseline measures (K. A. Lee et al.,2000) and non-pregnant women (Hertz et al., 1992; Schorret al., 1998); however, no changes across trimester (Brunneret al., 1994) and even an increase in SWS in pregnancy has beendocumented (Driver & Shapiro, 1992).

The mechanism of memory reprocessing during sleep isfar from understood. It is thought that during sleep, a centralmechanism for memory consolidation is the covert reactivationof neuronal populations used for encoding the respectivematerials during prior learning. The hippocampal replay ofpreviously encoded events drives a transfer of informationto the neocortex in which the memory becomes consolidatedand integrated into long-term representations (McClelland,McNaughton, & O’Reilly, 1995; Stickgold, Hobson, Fosse, &Fosse, 2001; Wilson & McNaughton, 1994). Different sleepstages are thought to contribute to consolidation of memoryin different ways. The dual-process hypothesis argues thatSWS facilitates the consolidation of declarative memory,whereas REM sleep facilitates the consolidation of implicitmemory (Plihal & Born, 1997). Alternatively, the sequential(two-step) model contends that SWS and REM sleep each playcomplementary roles and/or act serially to consolidate thememory trace (Ficca, Lombardo, Rossi, & Salzarulo, 2000;Giuditta et al., 1995; Stickgold, Whidbee, Schirmer, Patel, &Hobson, 2000).

When assessing the role of sleep in memory consolidationduring pregnancy, there are many potential confoundingfactors to consider. First, changes in respiratory function duringpregnancy can lead to alterations in maternal oxygenationduring sleep (Connolly et al., 2001; Prodromakis, Trakada, Tsa-panos, & Spiropoulos, 2004). Hypoxic episodes during sleep canbe associated with damage to the brain if they frequently recur(Gibson, Pulsinelli, Blass, & Duffy, 1981), and the hippocampusis particularly vulnerable to oxygen deprivation (Caine &Watson, 2000). Second, the role of attention is an importantconsideration. The clearest effect of sleep loss is sleepiness(Bonnet, 2000), which has been related to response slowingand attentional lapses (Durmer & Dinges, 2005). Lapses inattention decrease the ability of the person to focus and givenecessary effort to complete a task successfully. In addition,depressive symptomatology in the prenatal period is common(Andersson et al., 2003; A. M. Lee et al., 2007) and has shownto impair memory function (Sweeney, Kmiec, & Kupfer, 2000).Pregnant women identified as being depressed report poorersleep quality (Field et al., 2007; Jomeen & Martin, 2007), andsleep deprivation is a reliable predictor of both prenatal (Skou-teris, Germano, Wertheim, Paxton, & Milgrom, 2008) and post-natal mood changes (Wolfson, Crowley, Anwer, & Bassett,2003).

The purpose of this study was to examine the relationshipbetween sleep and memory during pregnancy, by establishingwhether sleep disturbances have any impact on memory

impairment in pregnancy while accounting for the poten-tial confounders of attention, mood, hormone level, andhypoxaemia during sleep. It was hypothesised that pregnantwomen would perform more poorly on episodic and proceduralmemory tasks in comparison with non-pregnant women, andthat this would be associated with measures of sleep disruption.The study also investigated the relationship between episodicand procedural memory and sleep stages, testing both the dual-process and the sequential hypotheses of memory consolidationduring sleep.

Method

Participants

Twenty-six women in the third trimester of pregnancy (T3:30–38 weeks gestation), 20 women in the first trimester ofpregnancy (T1: 9–14 weeks gestation), and 24 non-pregnantwomen (control group) participated in the study. The HumanResearch Ethics Committees at each institution involved inthis research approved this study, and informed consent wasobtained from all participants. Four hundred and thirty pregnantwomen from the Outpatient Obstetrics Clinic of a major mater-nity hospital were consecutively approached to participate in thestudy, and of these 56 agreed. The main reasons for decliningparticipation was an inability to be away from home overnightdue to family responsibilities and merely being unwilling to takepart in a sleep study. After volunteering to participate, tenpregnant women withdrew prior to data collection due topregnancy-related complications or inability to schedule anappropriate night to attend the sleep laboratory. Non-pregnantwomen were recruited from advertisements in the hospitalnewsletter and from friends of the pregnant participants. Partici-pants were excluded if they had a multiple or complicatedpregnancy, a significant medical, psychological, or psychiatricco-morbidity, a previously diagnosed sleep disorder, or a historyof head injury or memory problems. Uncorrected hearing, visualimpairment, poor English language skills, and antidepressantmedication use were also exclusions from participation.

Materials

The demographic information collected included age, handed-ness, relationship status, number of children, level of education,and employment status. Gestation in number of weeks was alsocollected for the pregnant participants. The instruments admin-istered include:• Depression Anxiety Stress Scale—Short version (DASS21;

Lovibond & Lovibond, 1995).• Wechsler Abbreviated Scale of Intelligence (WASI; The Psy-

chological Corporation, 1999).• Test of Memory Malingering (TOMM; Tombaugh, 1996). Par-

ticipants in this study were deemed to be giving sufficienteffort if they scored at least 45 of 50 (90%) on the secondtrial.

• Wechsler Memory Scale—Third Edition (WMS-III; Wechsler,1997). The raw scores from the four primary subtests oflogical memory, faces, verbal-paired associates, and familypictures were used to assess episodic memory. The Rarely

DL Wilson et al. Memory consolidation and sleep in pregnancy


197

Missed Index for logical memory recognition (Killgore &DellaPietra, 2000) was also calculated as a measure ofresponse validity.

• Rey Auditory-Verbal Learning Test (RAVLT; Rey, 1964).Memory retention was defined as the number of wordsretained on delayed recall as a percentage of the number ofwords recalled on the fifth immediate recall trial; therefore,a higher percentage indicates better memory retention.

• Austin (Milner) Maze (Milner, 1965; Walsh, 1985). Errors oneach of 10 trials were calculated as the number of times theparticipant presses a button that is not on the correct path.Delayed recall was assessed by the number of errors made onone further trial. Memory retention was defined as thenumber of errors made on the delay trial minus the numberof errors made on the tenth learning trial, resulting in a“difference” score.

• Motor-sequence learning. This procedural memory task isthe same as the finger-tapping task used by Walker, Brake-field, Morgan, Hobson, and Stickgold (2002), but for thisstudy will be termed motor-sequence learning so as not toconfuse with the finger-tapping test (Reitan, 1979) usedin neuropsychological batteries. Motor-sequence learningrequires participants to press four numeric keys on astandard computer keyboard with the fingers of their left(non-dominant) hand, repeating the five element sequence4-1-3-2-4 as quickly and as accurately as possible for 30 s.The numeric sequence was displayed at the top of the screenat all times to exclude any working memory component tothe task. Training consisted of ten 30-s trials with 30-s restperiods between trials. The scores (number of sequences anderrors) from the final two trials were averaged and taken asthe “post-training” performance. The averaged scores of twofurther 30-s trials assessed delayed performance. Memoryretention was defined as the number of sequences and errorsmade on delay minus those made on “post-training.” Apositive difference score therefore represents an increasednumber of sequences or errors after the retention interval.

• Mirror-tracing task (Model 31010; Lafayette InstrumentCo., Lafayette, IN, USA). For this task, participants wereinstructed to quickly and accurately trace a flat, six-pointedstar with a pencil while only a mirror-inverted image ofthe star was visible. Performance on each of ten trials wasassessed by the number of errors (drawing outside theedges of the star) and the time taken to trace the star.The averaged scores from the final two trials were taken asthe “post-training” performance, and the averaged scores oftwo further trials assessed delayed performance. Memoryretention was defined as the tracing time and errors made ondelay minus those made on “post-training”; a positive differ-ence score represents increased tracing time or errors afterthe retention interval.

• Test of variables of attention (TOVA; The TOVA Company,Los Alamitos, CA, USA). Variables measured includeresponse time, variability of response time (consistency),errors of commission (impulsivity), and errors of omission(inattention).

• PSG. Overnight PSG was conducted in laboratory withthe Somté (Compumedics, Abbotsford, Australia) portablesleep-monitoring device to control for variations in external

disruptions in the home environment and to provide greatercomfort for the participants. Polysomnogram recordings weresleep staged by a single experienced sleep technologist whowas blinded to pregnancy status, in accordance with standardcriteria (Rechtschaffen & Kales, 1968). Recent results froman Australian intra- and inter-laboratory scoring concordanceprogram showed the sleep technologist to have an agreementof 84.4–92.2% for individual sleep-stage scoring and 98.9%agreement for scoring of wake versus non-rapid eye move-ment (NREM) versus REM sleep. Sleep parameters includedtotal sleep time (TST) in minutes, number of minutes in stage1 sleep (NREM1), stage 2 sleep (NREM2), SWS (stages 3 and4 combined) and REM sleep, sleep latency, number of awak-enings during sleep, and wake after sleep onset (WASO) inminutes. Sleep cycle analysis was based on that used byMazzoni et al. (1999). Variables measured were the number ofsleep cycles, average sleep cycle length in minutes, and totalcycle time (TCT) as a proportion of TST (TCT/TST). Arousalsfrom sleep were measured in accordance with the rules setout by the American Sleep Disorders Association Atlas TaskForce (Bonnet et al., 1992). Lowest arterial oxygen saturationwas recorded, as well as percentage of TST with arterialoxygen saturation less than 95%.

Procedure

Participants arrived at the sleep laboratory in the evening,having refrained from drinking alcohol and caffeinated bever-ages from midday. The testing battery was administered as pre-sented in Table 1. All participants were tested at the same timeof day by the same investigator. The evening testing session tookapproximately 2 hours, and the participants were given restbreaks as requested.

On completion of testing, participants were set-up with theportable sleep-monitoring device and allowed to go to bed in aprivate room. Participants were left undisturbed until they werewoken 8 hours after lights out time. Approximately 30 min afterwaking, the delayed components of the memory tests were

Table 1 Schedule of Neuropsychological Testing and Estimated

Administration Time

Order of testing Administration

time (min)

Demographics <5

DASS21 <5

WASI 30

TOMM 10

Motor-sequence learning 10

Mirror-tracing task 10

Austin Maze 10

WMS-III 20

RAVLT 5

TOVA 22

Note. DASS21 = Depression Anxiety Stress Scale 21; WASI = Wechsler

Abbreviated Scale of Intelligence; TOMM = Test of Memory Malingering;

WMS-III = Wechsler Memory Scale—Third Edition; RAVLT = Rey Auditory

Verbal Learning Test; TOVA = Test of variables of attention.

Memory consolidation and sleep in pregnancy DL Wilson et al.


198

administered in the same order as in the evening, and bloodsamples were taken for serum progesterone levels due to itspotentially soporific effect (Söderpalm, Lindsey, Purdy, Hauger,& de Wit, 2004).

Statistical Analysis

All statistical analyses were performed with SPSS 15.0 (SPSSInc., Chicago, IL, USA). Data were checked for linearity andnormality, and non-normally distributed variables were eithersquare root or log transformed as appropriate. The few extremeunivariate outliers found (z score > 3.29) were assigned a rawscore one unit larger or smaller than the next most extremescore in the distribution, as recommended by Tabachnickand Fidell (2007). Data were screened for multivariate outliersusing Mahalanobis distance, and none was found. Four third-trimester women, one first-trimester woman, and one controlwoman were unable to complete the mirror-tracing taskbecause of its difficulty. TOVA data for one first-trimesterwoman were invalid because of an unanticipated disruptionduring the fourth quarter of the test. Up until this point, per-formance was within normal limits, and therefore this partici-pant was included in all other analyses.

One-way between-groups multivariate analysis of vari-ance (MANOVA) was used to compare groups on memoryretention on the WMS-III, the RAVLT, the Austin Maze,the procedural memory tests, sleep parameters, and moodstate. A mixed 3 (group) ¥ 4 (quarter) analysis of variance(ANOVA) was used to analyse TOVA response time andresponse time variability. To determine which variables con-tributed to memory retention during pregnancy, one-waybetween-groups analysis of covariance (ANCOVA) were alsoconducted, and the correlation matrix was examined to checkfor multicollinearity.

Effect sizes were calculated using eta squared and partialeta squared with 95% confidence intervals (CI) according toSmithson’s (2003) method. Effect sizes of 0.01, 0.06, and 0.14are considered small, medium, and large in magnitude, respec-tively. On those items where a significant difference betweengroups was detected, further investigation was conducted with

post hoc Tukey tests with the significance level set at p < .05.All values are given in mean � standard deviation (SD) fornormally distributed variables or median (Mdn) and inter-quartile range (IQR) for non-normally distributed variables.

Results

Participants

The average age in years of the third-trimester (M = 32.2,SD = 3.6), first-trimester (M = 29.4, SD = 3.3), and controlgroups (M = 29.3, SD = 5.9) did not differ (p = .09). Furtherdemographic details for each group are presented in Table 2. Thethree participant groups did not differ in terms of handedness,education level, employment status, or whether they alreadyhad children. However, significantly more pregnant womenwere in a stable relationship as compared with the controlgroup.

Intellectual Functioning

The third-trimester, first-trimester and control groups wereequivalent in terms of verbal and performance intelligence quo-tient (IQ), and the full-scale IQ did not differ across groups(control: M = 114.8, SD = 8.2; T1: M = 111.1, SD = 8.6; T3:M = 114.1, SD = 10.7; F[2, 67] = 0.96, p = .39).

Measures of Effort

All participants scored 45 or greater on the second trial of theTOMM, indicating that no participant was feigning memorydifficulties. Two control participants scored under the cut-off of136 on the Rarely Missed Index for logical memory recognition.However, both participants scored above average on generalmemory on the WMS-III and were deemed to be giving suffi-cient effort. All participants performed better on RAVLT recog-nition as compared with the first trial (Greiffenstein, Baker, &Gola, 1996) and delayed recall (Bernard, Houston, & Natoli,1993), further indicating adequate effort.

Table 2 Percentage of Participants in Each Demographic Category

Control (n = 24) T1 (n = 20) T3 (n = 26) c2 p

Right handed 87.5 95.0 96.2 1.60 .45

Relationship status 25.61** <.001

Married/de facto 37.5 85 92.3

Relationship 25 15 7.7

Single 37.5 0 0

Has children 25 55 46.2 4.43 .11

Tertiary educated 87.5 100 76.9 5.38 .07

Employment 8.20 .09

Full time 50 45 38.5

Part time 50 50 38.5

Unemployed 0 5 23

Note. Values given in percentages. T1 = first trimester of pregnancy; T3 = third trimester of pregnancy.

**p < .01.



199

Memory Retention

The means, SDs, significance levels, and effect sizes with 95% CIfor each between-group comparison on memory retentionscores for each memory test are presented in Table 3.

Episodic memory

Measures of verbal episodic memory retention (LogicalMemory, Verbal Paired Associates, and RAVLT) differed signifi-cantly across the groups, Wilks = 0.65, F(6, 130) = 5.26, p < .001,partial n2 = 0.20, (CI [0.06–0.28]). In particular, the first-trimester pregnant group retained significantly less informationon the Logical Memory task as compared with the third-trimester pregnant and control groups, and the effect size waslarge. The third-trimester pregnant group retained a signifi-cantly lower percentage of words from the RAVLT over theretention interval as compared with the control group, with alarge effect size. There were no differences across the groups onretention for the Verbal Paired Associates task.

Measures of visual episodic memory retention (Faces,Family Pictures, Austin Maze) did not differ across the groups,Wilks = 0.94, F(6, 130) = 0.73, p = .63, partial n2 = 0.03, (CI[0.00–0.06]). All groups retained close to 100% on the FamilyPictures task, and on average, participants performed better onthe Faces task after the retention interval. Neither of theseresults was due to ceiling effects on the tasks. On the AustinMaze, all groups showed an increase in errors made on delaycompared with the tenth trial of learning, and there was nodifference between the groups.

Procedural memory

On average, all groups improved to a similar degree over theretention interval on the Motor-sequence Learning task byincreasing the number of sequences and by decreasing thenumber of errors, and subsequently there was no significantdifference in performance across the groups. On the Mirror-tracing task, there was little change in tracing time over theretention interval. All groups decreased their error rate over theretention interval so that overall there was no significant differ-ence across groups on this task.

Associations With Verbal EpisodicMemory Retention

Sleep parameters

The means, SDs, significance levels, and effect sizes with 95% CIfor each between-group comparison for sleep parameters arepresented in Table 4. Key characteristics of sleep (TST, WASO,number of awakenings, and arousals/h) differed significantlyacross the groups, Wilks = 0.69, F(8, 128) = 3.34, p = .002,partial n2 = 0.17, (CI [0.03–0.24]). Specifically, women in thethird trimester of pregnancy had less TST than the controls andmore cortical arousals per hour than both the first trimestersand controls. WASO was significantly less in the control groupas compared with both pregnant groups. The effect sizes forthese differences were all medium to large. Sleep latency,or the time taken to fall asleep, did not differ across the groups.The overall difference in time spent in each sleep stageacross the groups demonstrated a strong trend, Wilks = 0.80,

Table 3 Mean (�SD) of Memory Retention Scores for Each Group and Level of Significance and Effect Sizes for Between-group Comparisons

Subtest Control (n = 24) T1 (n = 20) T3 (n = 26) F p Partial n2 (95% CI)a

Episodic tasks

LM % 82.4 � 11.4a 66.2 � 18.1b 79.1 � 15.1a 7.02b** .002 0.17 [0.03, 0.32]

VPA % 99.5 � 4.9 93.4 � 12.9 94.2 � 13.5 2.05b .14 0.06 [0.00, 0.17]

Faces % 106.3 � 8.4 103.4 � 10.0 105.3 � 8.1 0.60b .55 0.02 [0.00, 0.10]

FP %e 100.0 (98.1–103.0) 100.0 (97.7–101.2) 98.4 (95.6–100.0) 0.84b .44 0.02 [0.00, 0.11]

RAVLT % 87.7 � 13.0a 78.4 � 18.9ab 72.3 � 16.2b 5.79b** .005 0.15 [0.02, 0.29]

Maze 1.2 � 1.4 1.8 � 1.6 1.2 � 2.1 0.70b .50 0.02 [0.00, 0.11]

Procedural tasks

MS seq. 3.9 � 3.6 3.5 � 1.7 2.7 � 2.0 1.48c .24 0.04 [0.00, 0.15]

MS seq. % 17.7 � 19.0 17.4 � 10.3 13.2 � 11.3 0.73c .49 0.02 [0.00, 0.11]

MS errors -1.2 � 1.9 -0.9 � 1.7 -0.5 � 1.2 1.28c .29 0.04 [0.00, 0.15]

MT time sec. -2.2 � 5.4 0.8 � 5.0 -2.6 � 6.3 2.17d .12 0.07 [0.00, 0.20]

MT time % -5.4 � 16.1 2.7 � 14.3 -4.9 � 17.3 1.63d .20 0.05 [0.00, 0.17]

MT errors -5.0 � 5.5 -3.2 � 3.7 -3.5 � 2.9 1.08d .35 0.03 [0.00, 0.14]

Note. Means in the same row that do not share subscripts differ at p < .05 in the Tukey significant difference comparison. LM = Logical Memory; VPA = Verbal

Paired Associates; FP = Family Pictures; RAVLT = Rey Auditory Verbal Learning Test; Maze = Austin Maze errors; MS seq. = Motor-sequence Learning number

of sequences; MS seq. % = Motor-sequence Learning retention score as a percentage; MT time = Mirror-tracing; MT time % = Mirror-tracing time retention score

as a percentage; SD = standard deviation.a95% confidence intervals given in brackets (upper and lower).bdf = 2, 67.cdf = 2, 66.ddf = 2, 61.eValues given as Mdn (IQR) as variable was transformed.

**p < .01, *p < .05.



200

F(8, 128) = 1.94, p = .059, partial n2 = 0.11, (CI [0.00–0.16]).However, one-way ANOVA revealed that third-trimester preg-nant women spent significantly less time in REM sleep and SWSwhen compared with the controls. The effect sizes of thesedifferences were medium to large. In terms of sleep cycle archi-tecture, only the amount of TST spent in sleep cycles differedacross the groups, with the controls having a higher TCT/TSTpercentage than the pregnant women. The number of sleepcycles and the average sleep cycle length did not differ acrossthe groups. Measures of overnight arterial oxygen saturation didnot differ across the groups.

Attention

The average number of omission and commission errors madeby each group did not differ on any quarter of the attention test.Similar results were seen for total omission errors (control:M = 0.8, SD = 2.5; T1: M = 0.4, SD = 0.8; T3: M = 0.5, SD = 0.7)and total commission errors (control: M = 8.5, SD = 6.2;T1: M = 9.9, SD = 8.0; T3: M = 11.7, SD = 8.7), Wilks = 0.93,F(4, 130) = 1.15, p = .34. There was no interaction effectbetween the three groups and test quarter on the TOVAfor response time, Wilks = 0.98, F(6, 128) = 0.21, p = .97, andresponse time variability, Wilks = 0.93, F(6, 128) = 0.80, p = .57.A non-significant between-subjects effect indicates that thegroups did not differ overall on response time, F(2, 66) = 0.98,p = .31, or response time variability, F(2, 66) = 1.36, p = .26.

Progesterone

Progesterone level was significantly higher in the third-trimesterpregnant group (Mdn = 413 nmol/L, IQR = 311.5–509.3) com-

pared with the first-trimester pregnant group (68.5, 58.6–82.3),which was significantly higher than the control group (4.2,2.6–7.9; F(2, 65) = 392.38, p < .001, n2 = 0.92).

Mood state

The average DASS21 scores are shown in Table 5. Mood did notdiffer across the groups, Wilks = 0.87, F(6, 128) = 1.52, p = .18,n2 = 0.07, (CI [0.00–0.12]).

Associations Between Verbal Episodic MemoryRetention and Sleep During Pregnancy

The correlation matrix for memory retention, sleep parameters,and mood state for all participants is shown in Table 6. Asprogesterone is strongly linked to the progression of pregnancy,it could not be used in further covariate analyses so its influenceon memory retention needs to be investigated separately foreach group.

Table 4 Mean (�SD) of Sleep Parameters for Each Group and Level of Significance and Effect Sizes for Between-group Comparisons

Subtest Control (n = 24) T1 (n = 20) T3 (n = 26) Fa p Partial n2b (95% CI)

TST (min) 427.2 � 42.4a 403.7 � 38.4ab 388.6 � 55.9b 4.28* .018 0.11 [0.00, 0.25]

WASO (min) 28.0 � 19.5a 51.0 � 35.6b 59.1 � 33.8b 6.95** .002 0.17 [0.03, 0.31]

Awakenings 15.3 � 4.1 16.2 � 4.7 18.6 � 6.9 2.55 .09 0.07 [0.00, 0.19]

Arousals/hc 8.8 (7.1–10.9)a 10.6 (7.4–14.0)a 14.5 (10.6–18.2)b 10.46** <.001 0.24 [0.00, 0.38]

SL (min) 17.0 � 19.1 20.7 � 16.5 17.8 � 12.9 0.31 .73 0.01 [0.00, 0.07]

NREM1 (min) 27.8 � 12.2 29.4 � 15.6 35.6 � 15.6 2.00 .14 0.06 [0.00, 0.17]

NREM2 (min) 167.1 � 39.3 163.5 � 43.5 166.2 � 34.4 0.05 .95 0.001 [0.00, 0.03]

SWS (min) 157.5 � 39.5a 147.4 � 61.4ab 123.2 � 42.5b 3.40* .04 0.09 [0.00, 0.22]

REM (min) 75.0 � 18.1a 63.4 � 15.2ab 62.1 � 20.3b 3.65* .03 0.10 [0.00, 0.23]

Cycles (n.) 4.5 � 1.3 4.2 � 1.5 3.8 � 1.0 2.29 .11 0.06 [0.00, 0.18]

Cycles (min) 82.4 � 22.1 74.6 � 18.6 70.3 � 21.0 2.16 .12 0.06 [0.00, 0.18]

TCT/TST (%) 83.9 � 17.1a 73.1 � 14.3b 66.9 � 21.2b 5.64** .005 0.14 [0.01, 0.28]

Min O2 (%) 92.0 � 1.5 92.5 � 1.9 92.3 � 1.8 0.59 .56 0.02 [0.00, 0.10]

%TST O2 <95% 8.2 � 16.0 11.2 � 16.5 18.3 � 24.5 1.73 .19 0.05 [0.00, 0.16]

Note. Means in the same row that do not share subscripts differ at p < .05 in the Tukey significant difference comparison. TST = total sleep time; WASO = wake

after sleep onset; REM = rapid eye movement; SL = sleep latency; SWS = slow wave sleep; O2 = oxygen saturation; TCT = total cycle time; TST = total sleep

time.adf = 2, 67.b95% confidence intervals given in brackets (upper and lower).cValues given as Mdn (IQR) as variable was transformed.

**p < .01, *p < .05.

Table 5 Median (Interquartile) Scores on the DASS21 for Each Group and

Level of Significance and Effect Sizes for Between-group Comparisons

— Control T1 T3 Fa p n2

Depression 2 (0–4) 2 (2–5.5) 2 (0–4) 0.43 .65 0.01

Anxiety 1 (0–4) 2 (0–4) 4 (0–6) 1.72 .19 0.05

Stressb 9.6 � 5.9 8.4 � 7.9 8.3 � 6.3 0.26 .77 0.01

Note. Normal range—Depression = 0–9, Anxiety = 0–7, Stress = 0–14.

DASS21 = Depression Anxiety Stress Scale 21.adf = 2, 67.bValues given as M � SD.



201

Tab

le6

Inte

rcor

rela

tions

Bet

wee

nM

emor

yR

eten

tion

Scor

es,S

leep

Para

met

ers,

and

Moo

dSt

ate

—1

23

4a5

67

89

1011

1213

1415

1617

1819

2021

a22

a23

1.LM

0.17

0.33

**-0

.07

0.50

**0.

040.

05-0

.02

-0.1

10.

030.

110.

14-0

.15

0.24

*0.

10-0

.03

0.19

0.18

-0.1

00.

09-0

.08

-0.1

0-0

.06

2.V

PA-0

.05

-0.0

60.

240.

080.

120.

110.

120.

040.

230.

13-0

.32*

*0.

20-0

.07

0.03

0.05

0.10

0.15

0.29

*-0

.14

-0.2

5*-0

.11

3.Fa

ces

0.09

0.07

-0.0

7-0

.18

0.02

-0.0

50.

13-0

.09

0.08

0.02

0.05

0.07

0.03

-0.0

70.

04-0

.17

-0.1

3-0

.10

-0.0

1-0

.08

4.Fa

mP

ica

-0.0

80.

17-0

.09

0.18

-0.0

03-0

.06

-0.1

4-0

.07

0.21

-0.0

90.

090.

02-0

.05

-0.2

0-0

.04

-0.2

7*0.

130.

28*

-0.0

035.

RA

VLT

0.17

0.07

-0.1

60.

180.

040.

030.

09-0

.10

0.22

0.07

-0.1

00.

080.

150.

040.

19-0

.003

-0.0

80.

156.

Maz

e0.

07-0

.01

0.16

0.01

-0.0

2-0

.25*

0.09

-0.2

0-0

.19

0.18

-0.0

7-0

.28*

-0.0

70.

19-0

.16

0.02

-0.1

67.

MSL

seq

-0.2

9*-0

.08

-0.1

7-0

.01

-0.1

60.

090.

06-0

.02

0.04

0.02

0.06

-0.0

50.

050.

04-0

.08

-0.1

08.

MSL

erro

r-0

.003

-0.0

50.

020.

30*

-0.0

8-0

.08

0.13

-0.0

1-0

.06

0.05

-0.1

6-0

.13

-0.1

5-0

.30*

-0.3

1*9.

MT

time

0.26

0.05

0.10

-0.1

0-0

.06

-0.0

30.

05-0

.02

-0.1

10.

04-0

.08

-0.0

4-0

.03

0.02

10.

MT

erro

r0.

100.

01-0

.06

0.13

0.04

-0.0

90.

03-0

.03

0.07

0.03

-0.3

4*-0

.07

-0.1

111

.N

REM

10.

16-0

.53*

*0.

07-0

.05

0.03

0.39

**0.

09-0

.12

-0.0

1-0

.07

-0.2

7*-0

.12

12.

NR

EM2

-0.5

0**

0.29

*0.

43**

-0.3

4**

-0.0

80.

44**

-0.0

80.

23-0

.07

-0.2

7*-0

.22

13.

SWS

(min

)0.

001

0.45

**-0

.34*

*-0

.19

-0.0

10.

27*

0.13

-0.0

70.

100.

0214

.R

EM(m

in)

0.62

**-0

.49*

*-0

.19

0.48

**0.

160.

49**

-0.1

2-0

.20

-0.1

515

.TS

T(m

in)

-0.7

7**

-0.1

70.

55**

0.23

0.50

**-0

.20

-0.2

9*-0

.24*

16.

WA

SO0.

33**

-0.4

2**

-0.3

2**

-0.5

0**

0.08

0.14

0.01

17.

Aw

aken

0.03

-0.2

2-1

1-0

.15

-0.1

2-0

.22

18.

Cyc

le(n

.)-0

.31*

*0.

55**

-0.1

0-0

.32*

*-0

.28*

19.

Cyc

le(m

in)

0.57

**-0

.14

-0.1

30.

1620

.TC

T/TS

T-0

.25*

-0.3

5**

-0.0

521

.D

epre

ssio

na0.

42**

0.42

**22

.A

nxie

tya

0.48

**23

.St

ress

M76

.595

.810

5.1

100

79.3

1.4

3.4

-0.8

-1.6

-3.9

31.2

165.

714

1.9

66.9

406.

146

.116

.84.

275

.774

.52

28.

8SD

16.2

11.3

8.7

96.8

–101

.617

.11.

72.

61.

65.

84.

214

.738

.349

.418

.949

.132

.75.

61.

321

.119

.20–

40–

46.

6

Not

e.n

=70

.LM

=Lo

gica

lMem

ory

rete

ntio

n;V

PA=

Ver

bal

Pair

edA

ssoc

iate

s;Fa

mP

ic=

Fam

ilyP

ictu

res;

RA

VLT

=R

eyA

udito

ry-V

erb

alLe

arni

ngTe

stre

tent

ion;

MSL

=M

otor

-seq

uenc

eLe

arni

ng;M

T=

Mir

ror-

trac

ing;

NR

EM1

=st

age

1sl

eep

;N

REM

2=

stag

e2

slee

p;T

ST=

tota

lsle

eptim

e;W

ASO

=w

ake

afte

rsl

eep

onse

t;SW

S=

slow

wav

esl

eep

;REM

=ra

pid

eye

mov

emen

t;TC

T=

tota

lcyc

letim

e;TS

T=

tota

lsle

eptim

e.a V

aria

ble

was

tran

sfor

med

,val

ues

give

nas

Md

n(IQ

R).

**p

<.0

1,*p

<.0

5.



202

Within the control group, progesterone was not associatedwith Logical Memory or RAVLT retention. After combining thethird- and first-trimester pregnant groups, progesterone didnot significantly explain any additional variance in logicalmemory retention, F(1, 42) = 0.52, p = .48, or RAVLT retention,F(1, 42) = 2.09, p = .16, over that explained by number of weeksgestation.

Logical memory retention

As shown in Table 6, logical memory retention was signifi-cantly but weakly associated with higher levels of REM sleep.Logical memory retention was not related to any othermeasure of sleep, attention, or mood. Correlation matricesfor each individual group revealed no other significantcorrelations.

To determine whether REM sleep impacts the relationshipbetween pregnancy and logical memory retention, a one-waybetween-groups ANCOVA was performed. After adjustmentby minutes in REM sleep, logical memory retention still dif-fered significantly across pregnancy groups, as summarisedin Table 7, with F(2, 66) = 6.07, p = .004, partial n2 = 0.16, (CI[0.02–0.30]). The adjusted marginal means displayed in

Table 8 show that the first-trimester group scored significantlyless on logical memory retention as compared with the controlgroup. There was no relationship between the number ofminutes spent in REM sleep and logical memory retentionscore, F(1, 66) = 2.51, p = .12, n2 = 0.04, (CI [0.00–0.16]). Theresults indicate that the time spent in REM sleep is not con-tributing to reduced logical memory retention in the first tri-mester of pregnancy.

RAVLT retention

As presented in Table 6, the correlation between RAVLT reten-tion and REM sleep demonstrated a strong trend (p = .06), withbetter retention on the RAVLT weakly related to higher levels ofREM sleep. RAVLT retention is not correlated with any othersleep measures, attention variables, or mood state, and corre-lation matrices for each pregnancy group revealed no othersignificant correlations.

To determine whether REM sleep has an impact on the rela-tionship between pregnancy and RAVLT retention, a one-waybetween-groups ANCOVA was performed. After adjustment forminutes in REM sleep, RAVLT retention still differed signifi-cantly across pregnancy groups, as summarised in Table 7, withF(2, 66) = 4.36, p = .02, partial n2 = 0.12, (CI [0.00–0.25]). Theadjusted marginal means displayed in Table 8 show thatthe third-trimester group scored significantly less on RAVLTretention as compared with the control group. There was norelationship between the number of minutes spent in REMsleep and RAVLT retention score, F(1, 66) = 0.99, p = .32, partialn2 = 0.02, (CI [0.00–0.11]). These results only differ slightly tothe one-way ANOVA result for RAVLT retention (see Table 3),indicating that controlling for REM sleep only marginallyimproved the mean scores for RAVLT retention.

Relationship Between Memory and Sleep

The correlation matrix presented in Table 6 enables testingof the dual process and sequential hypotheses of memory con-solidation during sleep. Logical memory retention and RAVLTretention were only correlated with minutes of REM sleep, andthese relationships were weak. Verbal Paired Associates reten-tion was significantly negatively correlated with minutes inSWS and had positive but weak relationships with REM andstage 1 sleep (p = .09 and p = .06, respectively). Family picturesretention showed a tendency towards a positive association withSWS (p = .08). However, due to the limited variability in VerbalPaired Associated and family pictures retention scores, thesecorrelations should be interpreted with caution. Increased stage2 sleep was significantly but weakly associated with improve-ment on errors on the Austin Maze, but a higher number oferrors on the motor-sequence learning task. Faces retention,motor-sequence learning number of sequences, and mirror-tracing time and number of errors was not correlated with anymeasures of sleep.

Spending a higher proportion of TST within sleep cycles wassignificantly related to higher scores on Verbal Paired Associatesand family pictures retention, but these associations were small.More sleep cycles was weakly associated with an improved errorrate on delay on the Austin Maze. Retention scores on Logical

Table 7 Analysis of Covariance of Logical Memory Retention and RAVLT

Retention

Source of variance Adjusted SS df MS F


Pregnant group 2,648.44 2 1324.22 6.07**

Covariate

REM (min) 547.58 1 547.58 2.51

Error 14,394.66 66 218.10

RAVLT retention

Pregnant group 2,247.98 2 1123.99 4.36*

Covariate

REM (min) 255.12 1 255.12 .99

Error 17,025.70 66 257.97

Note. RAVLT = Rey Auditory-Verbal Learning Test; SS = Sum of Squares;

REM = rapid eye movement; MS = mean square.

**p < .01, *p < .05.

Table 8 Adjusted and Unadjusted Mean Logical Memory Retention and

RAVLT Retention Scores for Each Pregnancy Group

Pregnancy group Adjusted mean Unadjusted mean


Control 81.1 82.4

T1 66.7 66.2

T3 79.9 79.1

RAVLT retention

Control 86.8 87.7

T1 78.8 78.4

T3 72.8 72.3

Note. T1 = first trimester of pregnancy; T3 = third trimester of pregnancy.

RAVLT = Rey Auditory-Verbal Learning Test.



203

Memory, Faces, RAVLT, Motor-sequence Learning or Mirror-tracing were unrelated to sleep cycle architecture.

Discussion

The results support the initial hypothesis of memory difficultiesduring pregnancy, with pregnant women performing morepoorly on tasks of verbal episodic memory retention but not ontasks of visual episodic or procedural memory retention. Inparticular, the first-trimester pregnant women retained lessinformation from the paragraph recall task, whereas the third-trimester women retained less information from the word-listtask, as compared with non-pregnant women. The magnitudeof the effect sizes for these differences in performance was large.

To explain reduced verbal memory retention during preg-nancy, a number of factors were measured. Attention, mood,and nocturnal arterial oxygen saturation did not differ across thepregnant groups, and none of these were associated with verbalmemory retention. Although progesterone level varied greatlyacross the pregnant groups, this hormone was also unable toexplain any variance in memory retention.

Our key hypothesis was that sleep disruption during preg-nancy would be a factor in reduced memory retention. Pregnantwomen, particularly those in the third trimester, had moredisrupted sleep compared with non-pregnant women. Specifi-cally, they spent less time asleep, had more cortical arousals, andspent less time in SWS and REM sleep. Both pregnant groupsspent more time awake during the night and less of their TSTwithin sleep cycles, indicating more fragmented sleep. Theamount of information retained on the paragraph recall taskand the word-list task was significantly but weakly related totime spent in REM sleep only. However, further analyses indi-cated that contrary to expectation, reduced REM sleep in bothpregnant groups did not account for differences in verbalepisodic memory retention.

In looking at the double-step hypothesis, this study did findweak associations between sleep stages and declarative andprocedural memory. In contrast to theoretical expectation,however, higher amounts of REM sleep but lesser amounts ofSWS were weakly related to improvements on tasks of verbalmemory. Those who performed more poorly on Verbal PairedAssociates and family pictures retention had a tendency tospend less of their sleep time in sleep cycles, lending potentialsupport to the sequential hypothesis of memory consolidationduring sleep. In sum, the small size correlations found betweenmeasures of memory and sleep do not ultimately favour any ofthe existing theories on sleep-dependent memory consolida-tion, but they do not allow complete rejection of them either,partly due to the limited discriminant properties of somememory tasks.

Our results contrast the previous experimental studies whichhave noted that explicit memory is enhanced after early sleepperiods dominated by SWS (Drosopoulos, Wagner, & Born,2005; Plihal & Born, 1997, 1999; Yaroush, Sullivan, & Ekstrand,1971), or those showing that implicit memory benefits fromsleep with high amounts of REM sleep in the later part of thenight (Plihal & Born, 1997, 1999). However, our results maygive some support to studies showing that consolidation ofepisodic memory is related to sleep architecture, or the structure

of sleep cycles during the night (Ficca et al., 2000; Mazzoniet al., 1999; Stickgold et al., 2000).

For the motor-sequence learning task, our results for thefirst-trimester pregnant and non-pregnant groups were compa-rable with previous studies showing overnight improvementsin speed of between 17% and 20% (Kuriyama, Stickgold, &Walker, 2004; Walker et al., 2002). The third-trimester pregnantgroup showed a lesser improvement of 13.2%. In contrast toother studies, however (Fischer, Hallschmid, Elsner, & Born,2002; Walker et al., 2002), we were unable to attribute over-night improvement in speed to either stage 2 sleep or REMsleep.

One possible reason for the discrepancy between our resultsand previous findings of a strong relationship between sleepand memory consolidation could be a key difference in experi-mental design. Most methodologies manipulate sleeping condi-tions to test their hypotheses, whereas our study compared agroup of “normal” sleepers against those who were expected tohave disrupted sleep. For example, by dividing the retentioninterval into either early or late night sleep, Plihal and Born(1997) found that time spent in SWS was five times longerduring the early than late sleep retention interval, and the timein REM sleep was twice as long during late than early sleep. Incomparison, the third-trimester pregnant women in this studyhad 34 min less SWS and 13 min less REM sleep on averagethan the non-pregnant women. It may be that sleep stage archi-tecture needs to be significantly altered before memory consoli-dation is affected. Conditions involving more severe disruptionof sleep architecture include obstructive sleep apnoea (OSA)and primary insomnia. Patients with OSA have shown tohave significant memory impairments following a night of frag-mented sleep (Daurat, Foret, Bret-Dibat, Fureix, & Tiberge,2008; Kloepfer et al., 2009); however, relative contributions ofsleep disruption versus intermittent hypoxaemia need to beconsidered. Primary insomnia patients have shown to displayless overnight consolidation of both declarative (Backhaus et al.,2006) and procedural memories (Nissen et al., 2011), suggestingthat the significantly disturbed sleep profiles of these individualsare impairing the sleep-related memory consolidation process.

On the other hand, recent findings show that even short sleepperiods may be sufficient to promote memory consolidation. Forexample, declarative and procedural memory performance hasshown to benefit from half a night of sleep (Tucker & Fishbein,2009), and declarative memory retention has shown to improveafter hour-long napping (Tucker et al., 2006) and even withshort 6-min naps, suggesting that the mere onset of sleep mayinitiate the active processes of memory consolidation (Lahl,Wispel, Willigens, & Pietrowsky, 2008).

Side effects of sleep deprivation may include emotional andattentional disorders, reduced motivation, and disturbances inbiological rhythms; these disturbances may affect behaviouralperformance (Rauchs, Desgranges, Foret, & Eustache, 2005).Our study attempted to control for as many contributing factorsas possible. Measures of mood, attention, and progesteronewere all taken, and none of these were related to memoryretention. It has been argued that women may subconsciouslyperform more poorly on testing due to cultural expectationsof cognitive decline during pregnancy (Crawley, Grant, &Hinshaw, 2008). Incorporating well-validated tests of effort into



204

our methodology revealed that all pregnant participants weregiving sufficient effort towards testing, and therefore reducedmotivation was unlikely to be a factor.

Progesterone is one of the most active hormones during preg-nancy, but oestradiol, testosterone, oxytocins, and cortisol alsochange dramatically (Buckwalter et al., 1999). Kinsley et al.(2006) showed that progesterone and oestradiol are capable ofaltering the concentration of dendritic spines in the CA1 regionof the female rat hippocampus, a key area involved in memoryconsolidation. In non-pregnant studies, fluctuations in circulat-ing endogenous or exogenous oestrogens have been associatedwith cognitive changes (Kampen & Sherwin, 1994; Sherwin,1994). Although our single measure of progesterone did notrelate to reduced memory performance, it may be that a morecomplex interplay of a combination of pregnancy hormones isresponsible.

A major problem facing this area of research is that the termssleep, memory, and consolidation all refer to complex phenom-ena, none of which can be treated as a singular event (Stickgold,2005). For example, our study classified sleep into stages andcycles. However, others have investigated more fine-grainedparameters of sleep in an attempt to relate sleep to memoryconsolidation. The intensity of REM sleep in terms of numberand density of REMs was shown to increase during the nightrather than actual time spent in REM sleep, following proce-dural task acquisition (Smith, Nixon, & Nader, 2004). Verbalmemory retention and procedural learning have each beenrelated to the number of sleep spindles overnight, but notthe amount of time spent in any sleep stage (Clemens, Fabo, &Halasz, 2005; Gais, Molle, Helms, & Born, 2002; Morin et al.,2008). Investigation of electroencephalogram spectral powerhas suggested that REM sleep theta activity is involved indeclarative memory consolidation (Fogel, Smith, & Cote, 2007).This vast range of findings suggests that brain plasticity duringsleep may not involve a unitary process.

Limitations

In our study, we used only one of many experimentalapproaches commonly used to test the memory consolidationduring sleep hypothesis. Our method of measuring only onenight of sleep did not allow us to record changes in sleepparameters compared with baseline following memory training;similarly, we were unable to tell how the night of sleep prior totesting impacted on the initial encoding process in the memorytesting session. An individual sleep study also did not allowmeasurement of any first night effect; however, previous PSGresearch during pregnancy found no differences in sleepcharacteristics when including an adaptation night (K. A. Leeet al., 2000).

The pregnant women in this study slept less and had poorersleep quality than the non-pregnant women, but their sleep wasrelatively preserved when compared with controlled laboratorystudies of the effects of sleep deprivation or sleep restriction,where sleep time is often limited to 4 h (Banks, Van Dongen,Maislin, & Dinges, 2010; Casement, Broussard, Mullington, &Press, 2006; Guilleminault et al., 2003). Given the relativelysmall differences in sleep parameters across pregnancy groups,we would only expect to find small effects on measures of

memory consolidation. A larger sample size with more powermay have been required to identify any relationship betweensleep and memory consolidation; however, the sample size ofour study was large enough to have an 80% chance of findingcorrelations as small as r = 0.35 at a significance level of 0.05.Unfortunately, recruitment difficulties due to the requirementof having an overnight sleep study during pregnancy preventedus from increasing the study’s power any further. The lowparticipation rate during recruitment may also draw into ques-tion the representativeness of our study population, but this isacknowledged as a limitation of research involving comprehen-sive objective measurement techniques that may be perceivedas time consuming or intrusive by the participant.

Although we controlled for levels of attention in this study, nomeasure was included to specifically control for levels of sleepi-ness. In hindsight, the addition of a questionnaire such asthe Stanford Sleepiness Scale or the Epworth Sleepiness Scalewould have allowed for quick and easy measurement of sub-jective sleepiness. The widely accepted standard method ofquantifying sleepiness is the multiple sleep latency test (MSLT;Carskadon et al., 1986, p. 519), which measures the speed offalling asleep. The MSLT is based on the assumption that sleepi-ness is a physiological need state that leads to an increasedtendency to fall asleep (Roehrs, 2011, pp. 45–46). Based on this,we can reasonably argue that as sleep latency, or time taken tofall asleep, did not differ across the pregnant groups, that objec-tive sleepiness was not a contributing factor to the differences inmemory performance.

An unanticipated limitation of this study relates to thememory tasks used. The main neuropsychological test used tomeasure episodic memory retention was the WMS-III, which iswidely known and commonly used for both research and clini-cal populations. On average the participants in this study wereof “above average” intelligence; however, on three of four ofthe WMS-III subtests, they averaged close to or above 100%retention, even after a delay period of 9 h (well exceeding therecommended 30-min delay for clinical purposes). This was notdue to a ceiling effect on the visually based subtests; these taskswere difficult enough to have discriminative potential at initialencoding, but once learned, the information was not at allforgotten. Since this study was conducted, the 4th edition of theWMS had been released (Wechsler, 2009), which has elimi-nated both the Faces and Family Pictures subtests in favour ofmore visually oriented tasks, with an increased focus on visualworking memory. These tasks may represent visual memoryability more purely as they allow less opportunity to applyverbal descriptions to the visual stimuli in order to aid recall.

Conclusions

The current study has demonstrated that pregnancy is associ-ated with reduced verbal episodic memory retention as com-pared with non-pregnant women, with visual and proceduralmemory remaining unaffected. Pregnant women also demon-strated significant disruption of sleep patterns, with less TST,more awakenings, less deep sleep, and less REM sleep. In con-trast to prevailing theories of memory consolidation duringsleep, reduced memory retention could not be attributed tosleep disruption in our sample. Furthermore, overnight memory



205

retention was unrelated to attention, mood, hormone level, ornocturnal arterial oxygen saturation.

Although our results do not lend strong support to hypoth-eses regarding memory consolidation during sleep, neither dothey discount them. It is possible that microstructual changes tosleep architecture not detected by standard PSG are responsiblefor the measured memory deficits, which may explain why nostrong associations between memory and sleep were found inthis study. Alternately, it may be that pregnant women in thisstudy had a sufficient amount of sleep to support interactionsbetween the hippocampus and neocortex to strengthen re-presentations during sleep, and that some other unmeasuredfactors resulted in reduced memory performance. The reasonbehind memory dysfunction during pregnancy still remains acontroversial topic, and further well-controlled research usingmore discriminant memory tasks is required to ascertainwhether other aspects of the sleep-related memory consolida-tion hypothesis can be applied to this phenomena.

Acknowledgements

The authors would like to thank the staff at the Austin HealthSleep Laboratory and the Mercy Hospital for Women Outpa-tients Department for their support of this project, and weappreciate the valuable contribution made by each of theresearch participants.

References

Andersson, L., Sundström-Poromaa, I., Bixo, M., Wulff, M., Bondestam, K.,& Åström, M. (2003). Point prevalence of psychiatric disorders duringthe second trimester of pregnancy: A population-based study.American Journal of Obstetrics and Gynecology, 189(1),148–154.

Backhaus, J., Junghanns, K., Born, J., Hohaus, K., Faasch, F., & Hohagen, F.(2006). Impaired declarative memory consolidation during sleep inpatients with primary insomnia: Influence of sleep architecture andnocturnal cortisol release. Biological Psychiatry, 60(12), 1324–1330.

Banks, S., Van Dongen, H. P., Maislin, G., & Dinges, D. F. (2010).Neurobehavioral dynamics following chronic sleep restriction:Dose-response effects of one night for recovery. Sleep, 33(8),1013–1026.

Bernard, L. C., Houston, W., & Natoli, L. (1993). Malingering onneuropsychological memory tests: Potential objective indicators.Journal of Clinical Psychology, 49, 45–53.

Bonnet, M. H. (2000). Sleep deprivation. In M. Kryger, T. Roth, & W. C.Dement (Eds.), Principles and practice of sleep medicine (3rd ed.,pp. 53–71). Philadelphia, PA: W. B. Saunders.

Bonnet, M. H., Carley, D., Carskadon, M. A., Easton, P., Guilleminault, C.,Harper, R. M., . . . Westbrook, P. (1992). EEG arousals: Scoring rulesand examples: A preliminary report from the Sleep Disorders AtlasTask Force of the American Sleep Disorders Association. Sleep, 15,173–184.

Brindle, P. M., Brown, M. W., Brown, J., Griffith, H. B., & Turner, G. M.(1991). Objective and subjective memory impairment in pregnancy.Psychological Medicine, 21, 647–653.

Brunner, D. P., Münch, M., Biedermann, K., Huch, R., Huch, A., & Borbély,A. A. (1994). Changes in sleep and sleep electroencephalogram duringpregnancy. Sleep, 17(7), 576–582.

Buckwalter, J. G., Stanczyk, F. Z., McCleary, C. A., Bluestein, B. W.,Buckwalter, D. K., Rankin, K. P., . . . Goodwin, T. M. (1999). Pregnancy,

the postpartum, and steroid hormones: Effects on cognition andmood. Psychoneuroendocrinology, 24, 69–84.

Caine, D., & Watson, J. D. (2000). Neuropsychological andneuropathological sequelae of cerebral anoxia: A critical review.Journal of the International Neuropsychological Society, 6, 86–99.

Carskadon, M. A., Dement, W. C., Mitler, M. M., Roth, T., Westbrook,P. R., & Keenan, S. (1986). Guidelines for the multiple sleep latencytest (MSLT): A standard measure of sleepiness. Sleep, 9, 519–524.

Casement, M. D., Broussard, J. L., Mullington, J. M., & Press, D. Z. (2006).The contribution of sleep to improvements in working memoryscanning speed: A study of prolonged sleep restriction. BiologicalPsychology, 72(2), 208–212.

Casey, P., Huntsdale, C., Angus, G., & Janes, C. (1999). Memory inpregnancy. II: Implicit, incidental, explicit, semantic, short-term,working and prospective memory in primigravid, multigravid andpostpartum women. Journal of Psychosomatic Obstetrics andGynaecology, 20, 158–164.

Christensen, H., Leach, L. S., & Mackinnon, A. (2010). Cognition inpregnancy and motherhood: Prospective cohort study. British JournalOf Psychiatry, 196, 126–132.

Christensen, H., Poyser, C., Pollitt, P., & Cubis, J. (1999). Pregnancy mayconfer a selective cognitive advantage. Journal of Reproductive andInfant Psychology, 17(1), 7–25.

Clemens, Z., Fabo, D., & Halasz, P. (2005). Overnight verbal memoryretention correlates with the number of sleep spindles. Neuroscience,132(2), 529–535.

Connolly, G., Razak, A. R. A., Hayanga, A., Russell, A., McKenna, P., &McNicholas, W. T. (2001). Inspiratory flow limitation during sleep inpre-eclampsia: Comparison with normal pregnant and nonpregnantwomen. European Respiratory Journal, 18, 672–676.

Crawley, R. A., Grant, S., & Hinshaw, K. (2008). Cognitive changes inpregnancy: Mild decline or societal stereotype? Applied CognitivePsychology, 22, 1142–1162.

Daurat, A., Foret, J., Bret-Dibat, J. L., Fureix, C., & Tiberge, M. (2008).Spatial and temporal memories are affected by sleep fragmentationin obstructive sleep apnea syndrome. Journal of Clinical andExperimental Neuropsychology, 30(1), 91–101.

de Groot, R. H. M., Hornstra, G., Roozendaal, N., & Jolles, J. (2003).Memory performance, but not information processing speed, may bereduced during early pregnancy. Journal of Clinical and ExperimentalNeuropsychology, 25(4), 482–488.

de Groot, R. H. M., Vuurman, E. F. P. M., Hornstra, G., & Jolles, J. (2006).Differences in cognitive performance during pregnancy and earlymotherhood. Psychological Medicine, 36, 1023–1032.

Driver, H. S., & Shapiro, C. M. (1992). A longitudinal study of sleep stagesin young women during pregnancy and postpartum. Sleep, 15(5),449–453.

Drosopoulos, S., Wagner, U., & Born, J. (2005). Sleep enhances explicitrecollection in recognition memory. Learning & Memory, 12, 44–51.

Durmer, J. S., & Dinges, D. F. (2005). Neurocognitive consequences ofsleep deprivation. Seminars in Neurology, 25(1), 117–129.

Ficca, G., Lombardo, P., Rossi, L., & Salzarulo, P. (2000). Morning recall ofverbal material depends on prior sleep organization. BehaviouralBrain Research, 112, 159–163.

Field, T., Diego, M., Hernandez-Reif, M., Figueiredo, B., Schanberg, S., &Kuhn, C. (2007). Sleep disturbances in depressed pregnant womenand their newborns. Infant Behavior and Development, 30, 127–133.

Fischer, S., Hallschmid, M., Elsner, A. L., & Born, J. (2002). Sleep formsmemory for finger skills. Proceedings of the National Academy ofSciences of the United States of America, 99(18), 11987–11991.

Fogel, S. M., Smith, C., & Cote, K. A. (2007). Dissociablelearning-dependent changes in REM and non-REM sleep in declarativeand procedural memory systems. Behavioural Brain Research, 180,48–61.



206

Gais, S., Molle, M., Helms, K., & Born, J. (2002). Learning-dependentincreases in sleep spindle density. Journal of Neuroscience, 22(15),6830–6834.

Gibson, G. E., Pulsinelli, W., Blass, J. P., & Duffy, T. E. (1981). Braindysfunction in mild to moderate hypoxia. American Journal ofMedicine, 70, 1247–1254.

Giuditta, A., Ambrosini, M. V., Montagnese, P., Mandile, P., Cotugno, M.,Grassi Zucconi, G., & Vescia, S. (1995). The sequential hypothesis ofthe function of sleep. Behavioral Brain Research, 69(1-2), 157–166.

Greiffenstein, M. F., Baker, J., & Gola, T. (1996). Comparison of multiplescoring methods for Rey’s malingered amnesia measures. Archivesof Clinical Neuropsychology, 11(4), 283–293.

Guilleminault, C., Powell, N. B., Martinez, S., Kushida, C., Raffray, T.,Palombini, L., & Philip, P. (2003). Preliminary observations on theeffects of sleep time in a sleep restriction paradigm. Sleep Medicine,4(3), 177–184.

Hertz, G., Fast, A., Feinsilver, S. H., Albertario, C. L., Schulman, H., &Fein, A. M. (1992). Sleep in normal late pregnancy. Sleep, 15(3),246–251.

Janes, C., Casey, P., Huntsdale, C., & Angus, G. (1999). Memory inpregnancy. I: Subjective experiences and objective assessment ofimplicit, explicit and working memory in primigravid and primiparouswomen. Journal of Psychosomatic Obstetrics and Gynaecology, 20,80–87.

Jomeen, J., & Martin, C. R. (2007). Assessment and relationship of sleepquality to depression in early pregnancy. Journal of Reproductive andInfant Psychology, 25(1), 87–89.

Kampen, D. L., & Sherwin, B. B. (1994). Estrogen use and verbal memoryin healthy postmenopausal women. Obstetrics and Gynecology, 83(6),979–983.

Keenan, P. A., Yaldoo, D. T., Stress, M. E., Fuerst, D. R., & Ginsburg, K. A.(1998). Explicit memory in pregnant women. American Journal ofObstetrics and Gynecology, 179, 731–737.

Killgore, W. I. S., & DellaPietra, L. (2000). Using the WMS-III to detectmalingering: Empirical validation of the rarely missed index (RMI).Journal of Clinical and Experimental Neuropsychology, 22(6),761–771.

Kinsley, C. H., Trainer, R., Stafisso-Sandoz, G., Quadros, P., Marcus, L. K.,Hearon, C., . . . Lambert, K. G. (2006). Motherhood and the hormonesof pregnancy modify concentrations of hippocampal neuronaldendritic spines. Hormones and Behavior, 49, 131–142.

Kloepfer, C., Riemann, D., Nofzinger, E. A., Feige, B., Unterrainer, J.,O’Hara, R., . . . Nissen, C. (2009). Memory before and after sleep inpatients with moderate obstructive sleep apnea. Journal of ClinicalSleep Medicine, 5(6), 540–548.

Kuriyama, K., Stickgold, R., & Walker, M. P. (2004). Sleep-dependentlearning and motor-skill complexity. Learning and Memory, 11(6),705–713.

Lahl, O., Wispel, C., Willigens, B., & Pietrowsky, R. (2008). An ultra shortepisode of sleep is sufficient to promote declarative memoryperformance. Journal of Sleep Research, 17(1), 3–10.

Lee, A. M., Lam, S. K., Sze Mun Lau, S. M., Chong, C. S., Chui, H. W., &Fong, D. Y. (2007). Prevalence, course and risk factors for antenatalanxiety and depression. Obstetrics and Gynecology, 110(5),1102–1112.

Lee, K. A., Zaffke, M. E., & McEnany, G. (2000). Parity and sleep patternsduring and after pregnancy. Obstetrics and Gynecology, 95(1), 14–18.

Lovibond, S. H., & Lovibond, P. F. (1995). Manual for the depressionanxiety stress scales (2nd ed.). Sydney: Psychology Foundation.

Mazzoni, G., Gori, S., Formicola, G., Gneri, C., Massetani, R., Murri, L., &Salzarulo, P. (1999). Word recall correlates with sleep cycles in elderlysubjects. Journal of Sleep Research, 8(3), 185–188.

McClelland, J. L., McNaughton, B. L., & O’Reilly, R. C. (1995). Why thereare complementary learning systems in the hippocampus and

neocortex: Insights from the successes and failures of connectionistmodels of learning and memory. Psychological Review, 102(3),419–457.

Milner, B. (1965). Visually-guided maze learning in man: Effects ofbilateral hippocampal bilateral frontal, and bilateral cerebral lesions.Neuropsychologia, 3, 317–338.

Morin, A., Doyon, J., Dostie, V., Barakat, M., Hadj Tahar, A., Korman,M., . . . Carrier, J. (2008). Motor sequence learning increases sleepspindles and fast frequencies in post-training sleep. Sleep, 31(8),1149–1156.

Nissen, C., Kloepfer, C., Feige, B., Piosczyk, H., Spiegelhalder, K.,Voderholzer, U., & Riemann, D. (2011). Sleep-related memoryconsolidation in primary insomnia. Journal of Sleep Research, 20,129–136.

Parsons, C., & Redman, S. (1991). Self-reported cognitive changeduring pregnancy. Australian Journal of Advanced Nursing, 9(1),20–29.

Plihal, W., & Born, J. (1997). Effects of early and late nocturnal sleep ondeclarative and procedural memory. Journal of CognitiveNeuroscience, 9(4), 534–547.

Plihal, W., & Born, J. (1999). Effects of early and late nocturnal sleep onpriming and spatial memory. Psychophysiology, 36, 571–582.

Prodromakis, E., Trakada, G., Tsapanos, V., & Spiropoulos, K. (2004).Arterial oxygen tension during sleep in the third trimester ofpregnancy. Acta Obstetricia et Gynecologia Scandinavica, 83,159–164.

Rauchs, G., Desgranges, B., Foret, J., & Eustache, F. (2005). Therelationships between memory systems and sleep stages. Journal ofSleep Research, 14, 123–140.

Rechtschaffen, A., & Kales, A. (1968). A manual of standardizedterminology, techniques, and scoring system for sleep stages ofhuman subjects. Los Angeles, CA: Brain Information Service.

Reitan, R. (1979). Manual for administration of neuropsychological testbatteries for adults and children. Tucson, AZ: ReitanNeuropsychological Laboratory.

Rey, A. (1964). L’examen clinique on psychologie. Paris: PressesUniversitaires de France.

Schorr, S. J., Chawla, A., Devidas, M., Sullivan, C. A., Naef, R. W., III, &Morrison, J. C. (1998). Sleep patterns in pregnancy: A longitudinalstudy of polysomnography recordings during pregnancy. Journal ofPerinatology, 18, 427–430.

Sharp, K., Brindle, P. M., Brown, M. W., & Turner, G. M. (1993). Memoryloss during pregnancy. British Journal of Obstetrics and Gynaecology,100, 209–215.

Sherwin, B. B. (1994). Estrogenic effects on memory in women. Annals ofthe New York Academy of Sciences, 743, 213–230.

Silber, M., Almkvist, O., Larsson, B., & Uvnas-Moberg, K. (1990).Temporary peripartal impairment in memory and attention and itspossible relation to oxytocin concentration. Life Sciences, 47,57–65.

Skouteris, H., Germano, C., Wertheim, E. H., Paxton, S. J., & Milgrom, J.(2008). Sleep quality and depression during pregnancy: A prospectivestudy. Journal of Sleep Research, 17, 217–220.

Smith, C. T., Nixon, M. R., & Nader, R. S. (2004). Posttraining increases inREM sleep intensity implicate REM sleep in memory processing andprovide a biological marker of learning potential. Learning andMemory, 11, 714–719.

Smithson, M. (2003). Confidence intervals. Belmont, CA: Sage.Söderpalm, A. H., Lindsey, S., Purdy, R. H., Hauger, R., & de Wit, H.

(2004). Administration of progesterone produces mild sedative-likeeffects in men and women. Psychoneuroendocrinology, 29(3),339–354.

Stickgold, R. (2005). Sleep-dependent memory consolidation. Nature,437, 1272–1278.



207

Stickgold, R., Hobson, J. A., Fosse, R., & Fosse, M. (2001). Sleep, learning,and dreams: Off-line memory reprocessing. Science, 294, 1052–1057.

Stickgold, R., Whidbee, D., Schirmer, B., Patel, V., & Hobson, J. A. (2000).Visual discrimination task improvement: A multi-step processoccurring during sleep. Journal of Cognitive Neuroscience, 12(2),246–254.

Sweeney, J. A., Kmiec, J. A., & Kupfer, D. J. (2000). Neuropsychologicimpairments in bipolar and unipolar mood disorders on the CANTABneurocognitive battery. Biological Psychiatry, 48, 674–685.

Tabachnick, B. G., & Fidell, L. S. (2007). Using multivariate statistics (5thed.). Boston, MA: Pearson Education.

The Psychological Corporation. (1999). Wechsler abbreviated scale ofintelligence (WASI). San Antonio, TX: The Psychological Corporation.

Tombaugh, T. N. (1996). Test of memory malingering (TOMM). New York:Multi Health Systems.

Tucker, M. A., & Fishbein, W. (2009). The impact of sleep duration andsubject intelligence on declarative and motor memory performance:How much is enough? Journal of Sleep Research, 18(3), 304–312.

Tucker, M. A., Hirota, Y., Wamsley, E. J., Lau, H., Chaklader, A., & Fishbein,W. (2006). A daytime nap containing solely non-REM sleep enhances

declarative but not procedural memory. Neurobiology of Learning andMemory, 86, 241–247.

Walker, M. P., Brakefield, T., Morgan, A., Hobson, J. A., & Stickgold, R.(2002). Practice with sleep makes perfect: Sleep-dependent motor skilllearning. Neuron, 35, 205–211.

Walsh, K. (1985). Understanding brain damage. A primer ofneuropsychological evaluation. London: Churchill Livingstone.

Wechsler, D. (1997). Wechsler memory scale (3rd ed. manual). SanAntonio, TX: The Psychological Corporation.

Wechsler, D. (2009). Wechsler memory scale (4th ed.). San Antonio,TX: Pearson.

Wilson, M. A., & McNaughton, B. L. (1994). Reactivation of hippocampalensemble memories during sleep. Science, 265(5172), 676–679.

Wolfson, A. R., Crowley, S. J., Anwer, U., & Bassett, J. L. (2003). Changesin sleep patterns and depressive symptoms in first-time mothers: Lasttrimester to 1-year postpartum. Behavioral Sleep Medicine, 1(1),54–67.

Yaroush, R., Sullivan, M. J., & Ekstrand, B. R. (1971). Effect of sleep onmemory II: Differential effect of the first and second half of the night.Journal of Experimental Psychology, 88(3), 361–366.



208

reduced verbal memory retention is unrelated to sleep disturbance during pregnancy

Documents