Napping in older people ‘at risk’ of dementia: relationships with depression, cognition, medical burden and sleep quality


  • Nathan Cross,

    1. Healthy Brain Ageing Program, Brain & Mind Research Institute, The University of Sydney, Camperdown, NSW, Australia
    Search for more papers by this author
  • Zoe Terpening,

    1. Healthy Brain Ageing Program, Brain & Mind Research Institute, The University of Sydney, Camperdown, NSW, Australia
    Search for more papers by this author
  • Naomi L. Rogers,

    1. Concord Medical School, Concord Centre for Cardiometabolic Health in Psychosis, The University of Sydney, Sydney, NSW, Australia
    Search for more papers by this author
  • Shantel L. Duffy,

    1. Healthy Brain Ageing Program, Brain & Mind Research Institute, The University of Sydney, Camperdown, NSW, Australia
    Search for more papers by this author
  • Ian B. Hickie,

    1. Healthy Brain Ageing Program, Brain & Mind Research Institute, The University of Sydney, Camperdown, NSW, Australia
    Search for more papers by this author
  • Simon J.G. Lewis,

    1. Healthy Brain Ageing Program, Brain & Mind Research Institute, The University of Sydney, Camperdown, NSW, Australia
    Search for more papers by this author
  • Sharon L. Naismith

    Corresponding author
    1. Healthy Brain Ageing Program, Brain & Mind Research Institute, The University of Sydney, Camperdown, NSW, Australia
    • Correspondence

      Sharon L. Naismith, DPsych (Neuro), Healthy Brain Ageing Program, Brain and Mind Research Institute, 94 Mallett St, Camperdown, NSW 2050, Australia.

      Tel.: +612 9351 0781;

      fax: +612 9351 0551;


    Search for more papers by this author


Sleep disturbance is prevalent in older adults, particularly so in those at a greater risk of dementia. However, so far the clinical, medical and neuropsychological correlates of daytime sleep have not been examined. The aims of this study were to investigate the characteristics and effects of napping using actigraphy in older people, particularly in those ‘at risk’ of dementia. The study used actigraphy and sleep diaries to measure napping habits in 133 older adults ‘at risk’ of dementia (mean age = 65.5 years, SD = 8.4 years), who also underwent comprehensive medical, psychiatric and neuropsychological assessment. When defined by actigraphy, napping was present in 83.5% (111/133) of participants; however, duration and timing varied significantly among subjects. Nappers had significantly greater medical burden and body mass index, and higher rates of mild cognitive impairment. Longer and more frequent naps were associated with poorer cognitive functioning, as well as higher levels of depressive symptoms, while the timing of naps was associated with poorer nocturnal sleep quality (i.e. sleep latency and wake after sleep onset). This study highlights that in older adults ‘at risk’ of dementia, napping is associated with underlying neurobiological changes such as depression and cognition. Napping characteristics should be more routinely monitored in older individuals to elucidate their relationship with psychological and cognitive outcomes.


With advancing age, there are significant changes to sleep and the circadian system, such as higher fragmentation of sleep and increased awakenings, with a decrease in percentage of total sleep time spent in the deeper stages of slow-wave sleep (SWS; Floyd, 2002). Furthermore, in many individuals circadian phase advance is common, which can result in premature waking or a reduction in the total duration of sleep (Pandi-Perumal et al., 2010). It is thus not surprising that self-reported sleep difficulties occur in about 50–60% of adults aged 60 years and over (Almeida and Pfaff, 2005).

The prevalence of sleep disturbance is also quite pronounced in adults with dementia and Alzheimer's disease (AD), and is a significant burden for caregivers, as well as a chief cause of institutionalization (for review, see Bliwise, 2004). Patients with AD also show a high level of daytime sleepiness, which is often accompanied by increased daytime sleep and poor nocturnal sleep quality (Pat-Horenczyk et al., 1998; Vitiello and Prinz, 1989) – research has shown both of which correlate with disease severity and poorer scores on tests of cognition (Bonanni et al., 2005; Moe et al., 1995).

Indeed, over the last two decades, there has been increasing recognition of the integral role that sleep plays in cognitive functioning. Not only is sleep disturbance associated with poor ‘next-day’ cognitive functioning (Hagewoud et al., 2010), it is now understood that sleep plays a more direct role in brain plasticity and memory consolidation (Abel et al., 2013). Of significance, studies examining younger healthy adults suggest that napping can enhance memory consolidation and broader aspects of cognition (Tucker et al., 2006), even following restricted nocturnal sleep (Brooks and Lack, 2006; Tietzel and Lack, 2001).

The literature pertaining to the positive effects of napping for cognition has to-date been conducted in young healthy samples; however, these findings are of immense interest for clinical research seeking to enhance memory in older adults, specifically in those over the age of 50 years with a history of depression and/or concerns about cognition who are known to be ‘at risk’ of developing dementia (Clement et al., 2008; Glodzik-Sobanska et al., 2007; Naismith et al., 2012; Norton et al., 2014). In these populations, easily implemented, cost-effective interventions are urgently needed.

Older people ‘at risk’ of developing dementia have prominent sleep disturbance. In particular, over 60% of those with mild cognitive impairment (MCI) report having poor sleep (McKinnon et al., 2014), and this in turn relates to compromised neuropsychological functioning in the domains of memory and executive functioning (Naismith et al., 2010). Additionally, for older people with affective symptoms, sleep disturbance is related to depressive symptom severity, poor memory and executive functioning (Naismith et al., 2011a,b). However, it is unclear to what extent such clinical samples engage in napping behaviours, and how this may relate to nocturnal sleep quality and other clinical characteristics such as depression, medical co-morbidities and cognitive functioning.

While there is a dearth of research into the practice of napping in these ‘at-risk’ older adults, it seems that in the general older population napping has been related to medical co-morbidities and increased mortality, especially if naps are frequent and long (for review, see Ancoli-Israel and Martin, 2006; Dhand and Sohal, 2006). Additionally, fragmentation of circadian rest–activity rhythms in older adults has been associated with cognitive deficits (Oosterman et al., 2009). However, these findings are still inconclusive, have mostly been epidemiological, and not investigated in well-characterised clinical samples.

To the authors' knowledge, only one study has examined napping in those at risk of dementia, with findings suggesting that short (< 30 min), frequent (four times weekly) naps were associated with an 84% decreased risk for developing AD (Asada et al., 2000). Another study investigated the role of napping in cognitive decline over 10 years, with baseline napping again being found to share associations with a decreased risk of cognitive decline (Keage et al., 2012). However, both these studies involved self-reported measures of sleep quality and napping habits. It is difficult to completely understand the relationship between napping and cognition without objective measures of sleep; however, this is an area that is lacking in the literature. Indeed, reviews of napping in older adults have acknowledged the need for further research into this area, including the use of objective rather than self-reported measures of napping (Most et al., 2012; Vitiello, 2000).

In this study, we aimed to characterise the napping habits of a sample of older ‘at risk’ adults using actigraphy, a previously validated technique in both healthy people and patients with Parkinson's disease (Bolitho et al., 2013; Kanady et al., 2011). In particular, the aim was to determine the association between the timing and duration of napping behaviours, and the level of medical burden, depressive symptoms, cognitive functioning and nocturnal sleep quality.

Materials and methods


One-hundred and thirty-three adults aged over 50 years were recruited from the Healthy Brain Ageing Program, at the Brain & Mind Research Institute, Sydney, Australia. This early intervention research clinic provides detailed medical, psychiatric, neurological and neuropsychological assessments for people aged over 50 years with new-onset cognitive and/or mood disturbance. Exclusion criteria were: a dementia diagnosis or a Mini-Mental State Examination (MMSE) < 24 (Folstein et al., 1975); neurological disease (e.g. Parkinson's disease, epilepsy); psychosis; prior stroke or head injury (with loss of consciousness > 30 min); and inadequate English. This study was approved by the University of Sydney Institutional Ethics Committee, and all participants gave written informed consent prior to study participation.



All participants were assessed by a psychiatrist who recorded a full medical history, physical examination including body mass index (BMI), current medications, and an assessment of psychiatric history using the Structured Clinical Interview for Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (DSM-IV; First et al., 2007). This clinical assessment also included the 30-item Geriatric Depression Scale (GDS; Yesavage et al., 1982); the Social and Occupational Functional Assessment Scale (SOFAS; American Psychiatric Association, 1994); the Cumulative Illness Rating Scale – Geriatric Version (CIRS-G; Miller et al., 1992); and the Multivariable Apnoea Prediction Index (Maislin et al., 1995).


As previously reported (Duffy et al., 2014), a standardised test battery was administered by a Clinical Neuropsychologist. The test battery was chosen for its capacity to examine broad aspects of cognition relevant to ageing and dementia, with specific emphasis on processing speed [Trail Making Test – Part A (TMT-A), expressed as an age- and education-adjusted z-score; Tombaugh, 2004], verbal memory [the Logical Memory II (LM-II) subtest of the Wechsler Memory Scale-III, expressed as an age-scaled score (ASS); Wechsler, 1997b] and executive functioning [the Controlled Oral Word Association Test (COWAT), z-score; Benton et al., 1983; Trail Making Test – Part B (TMT-B) z-score; Tombaugh, 2004; and the Stroop subtest (Part 3) of the Delis Kaplan Executive Functioning System (ASS); Delis et al., 2001]. Premorbid intellectual functioning was estimated using the Wechsler Test of Adult Reading (intelligence quotient estimate; Wechsler, 1997a). The neuropsychological assessment was performed between the hours of 09:00 and 15:00 hours.

Following these assessments, a consensus panel of one psychiatrist and two neuropsychologists determined whether a patient met the criteria for MCI, based on established criteria (Winblad et al., 2004).

Sleep-wake assessment

Measures included actigraphy, sleep diary and self-report, as follows:

  1. Actigraphy: as described previously (Naismith et al., 2010). Following neuropsychological assessment, participants were asked to wear a wrist actigraph (Actiwatch Spectrum, Koninklijke Philips N.V.) on the non-dominant wrist over the subsequent 2-week period. Actigraphy rest intervals were calculated using Actiware 5.0 software and Actiwatch Firmware Version 01.01.0007 (Minimitter-Respironics, Bend, OR, USA) in conjunction with manual scoring. Analysis of actigraphy data was conducted by an experienced sleep technician overseen by a chronobiologist, and incorporated sleep diary data with actigraphy-recorded activity and light information. Nap scoring was conducted using actigraphy based upon protocols previously reported (Bolitho et al., 2013). For a more detailed account of the scoring protocol, please see the supplementary materials provided. Although it is acknowledged that actigraphy cannot truly estimate ‘sleep’ as opposed to activity within a ‘rest interval’, measurements obtained from actigraphy were referred to as ‘sleep’ for ease of interpretation, as previously reported (Naismith et al., 2010).
  2. Sleep diary: Participants were asked to complete concurrently a sleep diary, which required them to provide information on their sleep patterns, such as bedtimes and wake times, sleep latency and disturbance, as well as napping times and durations.
  3. Self-report sleep–wake measures: these included the Pittsburgh Sleep Quality Index (PSQI; overall sleep quality; Buysse et al., 1989); the Epworth Sleepiness Scale (ESS; daytime sleepiness; Johns, 1991); and the Horne–Östberg Morningness–Eveningness Questionnaire (MEQ; chronotype and circadian phase; Horne and Ostberg, 1976).

Statistical analysis

Data were analysed using the Statistical Package for the Social Sciences (version 20 for Mac). Comparisons between napping and non-napping groups were analysed using Students t-tests. Analysis of categorical variables utilized a χ2-test. Pearson's coefficients were used for all correlations unless otherwise stated. All analyses were two-tailed and used an alpha level of 0.05. When analysing the duration of naps, three values were deemed to be outliers and were curtailed to 200 min.


Participants (51.1% female) were aged between 51 and 89 years, with a mean age of 65.5 years [standard deviation (SD) = 8.4 years] and 13.7 years of formal education (SD = 3.3 years). In terms of other patient demographics, 96 participants (72.2%) were not currently working (retired or receiving disability benefits), with 37 (27.8%) still in some capacity of employment or voluntary work.

Of this sample of 133 older adults, 68 participants (51.1%) had a lifetime history of major depression, of which 21 (15.8% of total sample) met DSM-IV criteria for a current major depressive episode. Additionally, 85 (63.9%) met criteria for MCI (of which 32 had no lifetime depression history), and 33 were health-seeking for cognitive and specialist assessment but did not meet objective criteria for MCI or depressive illness. Forty-eight participants (36.1%) were taking an antidepressant, nine (6.8%) participants were taking a benzodiazepine to assist sleep, and one participant (0.7%) was taking a new-generation non-benzodiazepine hypnotic (Z-drug). Finally, 87 (65.4%) were alcohol consumers, with the average alcohol consumption being 5.3 standard drinks per week (SD = 7.9 standard drinks per week).

From the sample, 88.7% (118/133) of participants completed sleep diaries, and of these 77 (65.3%) reported napping. When evaluated according to actigraphy, napping was detected in 111 (83.5%) participants (χ2 = 24.0, df = 2, < 0.001).

Table 1 shows the characteristics of actigraphy-defined nappers and non-nappers. On average, nappers did not differ from non-nappers in terms of age, gender, work status, social and occupational functioning (SOFAS), or overall cognition (MMSE). However, compared with non-nappers, nappers did have a greater BMI and reported greater medical burden (CIRS-G). These findings did not statistically change when napping was defined according to sleep diary self-report (Table 1b, supplementary materials). Additionally, because the sample of nappers was substantially larger than the sample of non-nappers, all analyses were repeated with a 2 : 1 random sample, and the significance of the results did not change (Table 1c, supplementary materials).

Table 1. Characteristics of nappers versus non-nappers as defined by actigraphy
Demographic, clinical and sleep characteristicsNappers (= 111)Non-nappers (= 22)t-valueP
  1. Data are mean ± SD, or % (n).

  2. BMI, body mass index; CIRS, Cumulative Illness Rating Scale; ESS, Epworth Sleepiness Scale; GDS, Geriatric Depression Scale; MCI, mild cognitive impairment; MEQ, Morningness–Eveningness Questionnaire; MMSE, Mini-Mental State Examination; PSQI, Pittsburg Sleep Quality Index; SOFAS, Social and Occupational Functioning Assessment Scale; WASO, wake after sleep onset.

  3. a

    Equal variances not assumed in Student's t-test.

  4. b


  5. c

    Actigraphy measurements are based upon the nocturnal rest period.

  6. Bold P-values indicate significantly different scores between groups.

Clinical measurements
Gender, femaleb52.21% (59/113)50.00% (11/22)0.010.908
Age, years65.88 ± 8.3962.59 ± 8.42−1.690.094
Current major depressive episode15.31% (17/111)18.18% (4/22)0.330.739
Current antidepressant use39.60% (41/107)27.3% (6/22)−1.0220.314
GDS,/3010.76 ± 8.149.23 ± 7.70−0.890.376
SOFAS score,/10075.63 ± 15.2175.41 ± 12.15−0.060.949
Medical burden (CIRS)a5.46 ± 4.802.27 ± 1.86−5.24 < 0.001
BMI28.12 ± 7.4724.13 ± 3.05−4.12 < 0.001
MMSE,/3028.38 ± 1.6826.68 ± 1.320.730.470
MCI diagnosisb67.56% (75/111)45.45% (10/22)4.26 0.039
Wechsler Test of Adult Reading-Intelligence Quotient estimate111.16 ± 53.70106.00 ± 6.73−0.450.654
Apnoea risk (Multivariable Apnoea Index),/10.37 ± 0.250.29 ± 0.17−1.340.184
Sleep quality (PSQI),/216.90 ± 4.105.33 ± 4.53−1.580.118
Daytime sleepiness (ESS),/247.42 ± 4.784.70 ± 4.27−2.33 0.022
Chronotype (MEQ),/8658.95 ± 9.1363.80 ± 7.452.22 0.028
Sleep latency (diary)23.82 ± 19.7318.23 ± 12.98−0.870.389
Actigraphy measurementsc
Rest interval onset, h23:05 ± 1:1622:51 ± 0:59−0.800.425
Rest interval offset, h7:32 ± 1:0707:17 ± 0:54−0.950.346
Total sleep time, min451.64 ± 68.09456.83 ± 48.170.340.734
WASO, min56.63 ± 27.2949.39 ± 15.95−1.200.232
Sleep efficiency (%)88.99 ± 5.1490.34 ± 2.95−1.200.234

Of those who were classified to have napped based on actigraphy measures, the average number of days on which a nap occurred over a fortnight was 3.9 (SD = 3.1 days, range = 1–14 days, median = 3.0 days). The median time of the first nap was 14:22 hours, with nap time ranging from 07:26 hours to 22:31 hours. For each day where a nap was present, there was an average of 1.1 naps per day (SD = 0.3 naps per day, range = 1.0–2.0 naps per day). The average duration of naps was 72.5 min (SD = 37.4 min, range = 17.5–200.0 min, median = 61.2 min).

Associations between napping and subjective sleep quality

As shown in Table 1, there were no differences in self-reported sleep quality as measured by the PSQI, or in diary reports of sleep latency, between those who napped and those who did not. However, nappers had significantly higher levels of daytime sleepiness on the ESS. Additionally, nappers also significantly differed from non-nappers on MEQ scores. On average, non-nappers were considered ‘morning types’ and nappers in the ‘intermediate type’ range were considered neither morning nor evening; although, within nappers, no significant relationships were found between any self-reported measures of sleep quality and the average time, frequency or duration of naps measured by actigraphy.

Associations between napping and sleep–wake assessment

When comparing nappers and non-nappers, there were no differences between sleep-onset time, sleep-offset time, nocturnal sleep disturbance [wake after sleep onset (WASO)], sleep efficiency or total sleep time. However, when only the nappers (as defined by actigraphy) were examined (Table 2), longer nap durations were significantly associated with later sleep offset. In addition, longer nap duration related to a higher amount of variance in sleep onset (= 0.3, = 0.005), offset (= 0.3, = 0.007) and total sleep time (= 0.2, = 0.019). Finally, later nap times on average each day were also associated with later sleep onset (= 0.2, = 0.020) and WASO (= −0.2, = 0.031). When the sample was compared between those who had their first nap before 17:00 hours (= 87) and after 17:00 hours (= 22), a first nap after 17:00 hours was associated with a shorter time in bed (= −2.16, = 0.033) and less WASO (= 2.27, = 0.025).

Table 2. The relationships between duration and onset of naps with demographic, sleep, depression and cognitive variables in nappers (= 111)
 AgeClinical characteristicsActigraphy sleep measuresNeuropsychological testsa
Medical burden, CIRSBMIGDSCurrent major depressive episodeAntidepressant useMMSESleep onsetSleep offsetTotal sleep timeWASOSleep efficiencyPsycho-motor speed, TMT-AVerbal memory, LM-IIVerbal fluency, COWATSet shifting, TMT-BResponse inhibition, Stroop
  1. Statistics represent Pearson r correlations.

  2. BMI, body mass index; CIRS, Cumulative Illness Rating Scale; COWAT, Controlled Oral Word Association Test; GDS, Geriatric Depression Scale; LM-II, Logical Memory II; MMSE, Mini-Mental State Examination; TMT-A, Trail Making Task – Part A; TMT-B, Trail Making Task – Part B; WASO, wake after sleep onset.

  3. *< 0.05; **< 0.01.

  4. a

    Neuropsychological tests are presented as normative data.

Duration of naps−0.0250.0640.0410.256**0.0850.307**−0.1020.0490.221*0.0840.171−0.116−0.210*−0.245*−0.241*0.001−0.156
Naps per day−0.0760.193*0.209*0.195*−0.058−0.084−0.055−0.063−0.0380.023−0.0030.232*−0.0480.053−0.205*−0.099−0.019
Time of first nap−0.0930.103−0.055−0.1230.132−0.0170.0220.221*0.018−0.131−0.206*−0.1740.0980.0330.0350.1360.057
Total number of naps0.217*0.334**0.245*0.176*0.1560.141−0.081−0.083−0.0090.0400.0780.076−0.066−0.023−0.176−0.234*−0.080

Associations between napping, clinical features and depressive symptoms

As shown in Table 2, the number of naps per day had a significant positive correlation with BMI and medical burden (CIRS-G severity). Self-reported depressive symptoms (GDS) did not differ between those who napped and those who did not. However, in the subsample that did nap, depressive symptom severity was associated with longer duration of naps and a greater number of naps per day (Table 2). Additionally, there were significant differences between participants who were currently involved in some form of work (including voluntary) and those who were not, with those who were not currently working napping for longer (= 2.1, df = 106, = 0.035) and on more days (= 2.2, df = 82, = 0.032).

Participants meeting criteria for current major depression (= 17) did not differ from those who did not (= 94) in terms of time (= 0.6, df = 109, = 0.543) or duration of naps (= −0.9, df = 107, = 0.382), or the total number of naps or naps per day (= −1.4, df = 109, = 0.166). Participants who were taking an antidepressant medication (= 42) did not differ from those not taking medication (= 69) in regards to the average time of their first nap each day (= 1.0, df = 106, = 0.319) or the average number of naps per day (= −1.4, df = 107, = 0.887). However, those on antidepressant medication did demonstrate a greater number of days napping (= −2.4, df = 75, = 0.024) and had a longer average nap duration (= −3.1, df = 65, = 0.003). There were no significant relationships between alcohol intake and napping in this sample.

Associations between napping and neuropsychological data

The proportion of participants with MCI was significantly higher in those who napped compared with those who did not (Table 1). In the napping subsample, those with MCI (= 75) did not differ from those that were cognitively intact (= 35) in terms of duration of naps (= −1.0, df = 106, = 0.327), time of first nap (= 0.2, df = 108, = 0.844) or the number of naps per day (= −0.6, df = 108, = 0.570). However, those meeting MCI criteria napped on average on an extra day (= −2.1, df = 86, = 0.041). Further analysis showed that there were no significant differences in napping between those with amnestic and non-amnestic MCI (all > 0.050).

As shown in Fig. 1, for those who napped, longer nap durations correlated significantly with poorer performance on verbal memory (LM-II), processing speed (TMT-A) and verbal fluency (COWAT) tasks. Additionally, having a greater number of naps per day was also correlated with poorer verbal fluency, and a greater number of naps over the measurement period was associated with poorer mental flexibility (TMT-B). These findings were still significant after controlling for depressive symptom severity (GDS), age and gender; and were still significant when those participants who were taking benzodiazepines were removed from the sample. While there were no significant relationships between the time at which the first nap occurred and any neuropsychological measures, those who had a first nap before 17:00 hours had significantly lower scores on processing speed (TMT-A; = 2.0, = 0.044) and mental flexibility (TMT-B; = 2.0, = 0.045) than participants who had their first nap post-17:00 hours (Table 3).

Table 3. Comparisons of neuropsychological and actigraphy measures between those who napped pre- and post-17:00 hours
Demographic, clinical and sleep characteristicsPre-17:00 hours (n = 89)Post-17:00 hours (n = 22)t-valueP
  1. Data are mean ± SD.

  2. COWAT, Controlled Oral Word Association Test; LM-II, Logical Memory II; TMT-A, Trail Making Task – Part A; TMT-B, Trail Making Task – Part B; WASO, wake after sleep onset.

  3. a

    Actigraphy measurements are based upon the nocturnal rest period.

  4. b

    Equal variances not assumed in Student's t-test.

  5. Bold P-values indicate significantly different scores between groups.

Neuropsychological test
TMT-A (z-score)−0.11 ± 1.170.43 ± 0.782.04 0.044
LM-II (age-scaled score)10.82 ± 3.3711.33 ± 3.510.630.532
COWAT (z-score)−0.14 ± 0.930.29 ± 1.081.860.066
TMT-B (z-score)−0.52 ± 1.840.31 ± 0.952.03 0.045
Stroop test 3 (age-scaled score)9.83 ± 3.2511.00 ± 2.321.500.136
Actigraphy measurementsa
Rest interval onset (h)22:58 ± 1:1723:31 ± 1:051.810.073
Rest interval offset (h)b07:32 ± 1:1307:28 ± 0:37−0.360.718
Time in bed (min)514.11 ± 73.92477.56 ± 58.02−2.16 0.033
Total sleep time (min)455.44 ± 69.74435.75 ± 59.61−1.190.236
WASO (min)59.47 ± 27.6144.70 ± 22.84−2.27 0.025
Sleep efficiency (%)88.57 ± 5.1790.71 ± 4.79−1.730.236
Figure 1.

Scatterplots showing the association between average nap duration and scores of processing speed, verbal fluency and verbal memory. (TMT-A = Trail Making Test – Part A; COWAT = Controlled Oral Word Association Test; LM-II = Logical Memory test – Part II)


This study sought to determine the characteristics and clinical correlates of napping in older adults ‘at risk’ of dementia. Of significance, the current data show that, according to actigraphy, over 80% of such patients nap, demonstrating that napping is a common feature in this help-seeking patient group with a history of depression, MCI and/or subjective memory complaints. In addition, higher rates of napping were associated with greater medical burden and a higher BMI, as well as whether or not the participant was currently engaged in regular work.

Overall, in terms of napping practices, there was considerable heterogeneity in the current data, with some 28 (25.2%) participants napping on only 1 day per fortnight, and only 1 (0.9%) napping every day. Additionally, the timing and duration of napping varied considerably. While, on average, napping occurred within the post-prandial period (approximately 14:00 hours), this ranged from about 07:00 hours to 22:00 hours. Early naps may have an increased likelihood of containing rapid eye movement (REM) sleep due to the circadian drive for REM sleep as opposed to the homeostatic drive for SWS, depending on the circadian phase of the participants. While there were significant differences in the circadian phase between those who napped and those who did not, circadian phase as measured by the MEQ shared no relationship with the time of the first nap. In addition, the average duration of napping was approximately 60 min, but ranged from periods of less than 20 min to over 3 h. This indicates that, while common, napping habits in older people exhibit a great deal of variance both within and between individuals. It would be of interest to obtain further objective measurements (e.g. electroencephalography) in older adults who nap, in order to obtain what effect the timing and duration of naps may have on the distinct sleep architecture elements that constitute these sleep periods.

With regard to the relationships between napping and nocturnal sleep quality, later nap times were on average associated with later times of sleep onset and an increase in nocturnal awakenings (WASO). Approximately 20% of this sample had an average naptime that was post-17:00 hours and, importantly, participants whose nap time was post-17:00 hours spent a shorter time in bed than those who napped before 17:00 hours, and surprisingly also spent less time awake throughout the night. Whilst the average time of first nap shared no association with cognition, those whose first nap was after 17:00 hours correlated with higher scores on neuropsychological outcomes than those who had their first nap before 17:00 hours. However, these patients who first napped before 17:00 hours also on average napped significantly more frequently each day, and scored higher for medical burden (on CIRS), which may account for these findings.

Longer naps, which are thought to contain SWS, close to habitual bedtime could influence homeostatic sleep pressure (i.e. increasing the latency to SWS), and thus reduce the quality of sleep (Werth et al., 1996). However, in this study, even though SWS could not be measured, nap duration had little effect on nocturnal sleep disturbance, either self-reported (i.e. PSQI, sleep latency) or measured by actigraphy (i.e. WASO and sleep efficiency). Nevertheless, whilst nap duration was not associated with the average nocturnal total sleep time, it did significantly correlate with the variance in the total time, onset and offset times of the nocturnal sleep period. This may be more telling than the average total sleep time, as these measurements were all collected over a period of 14 days, and thus occasional longer naps could negatively impact upon specific nights of sleep without influencing the overall average of either duration of the nap or major sleep period. It could also be possible that inconsistent or variable sleep and/or wake times due to excessive napping may elicit negative effects on daytime cognition.

Alternatively, studies specifically examining older adults have shown that a nap of approximately 80 min duration has little effect on subsequent sleep, and it has been suggested that a daytime nap may actually compensate for the reduced duration of nocturnal sleep that occurs in older people (Campbell et al., 2005). However, these naps only occurred in the afternoon, and therefore cannot provide information surrounding the effects of naps that occur close to habitual bedtime.

With regard to cognition, the duration of naps was pertinent. Participants that napped for longer tended to have poorer performance on neuropsychological tasks associated with slower processing speed (TMT-A), poorer verbal fluency (COWAT) and verbal memory (LM-II). While these relationships were negative, no direction of causality from this naturalistic and cross-sectional study design could be ascertained. Rather, it is entirely plausible that common neurobiological networks underpin both cognitive dysfunction and excessive daytime sleepiness leading to napping, as seen in the current sample. From these results, future experimental research examining prospectively the effect of nap duration on cognition appears warranted.

In younger age groups, experimental studies suggest that the ideal nap duration is determined by the way in which benefits emerge over the post-nap period (Brooks and Lack, 2006). In healthy samples, cognitive benefits can be observed with naps as short as 10 min (Brooks and Lack, 2006), but 45 min naps tend to be favoured by older people and may improve alertness for periods up to 3 h later (Takahashi and Arito, 2000). In terms of the components of sleep microarchitecture that contribute to superior memory consolidation, studies to-date have varied, though are mostly attributed to enhancements to either SWS (Tucker et al., 2006), or delta power and/or sleep spindles in Stage 2 sleep (Wamsley et al., 2010). The use of an ‘unlimited’ non-rapid eye movement (NREM) period for napping (i.e. allowing for deep SWS) is supported by data showing that completion of the NREM sleep cycle (evidenced by the onset of REM sleep) may be critical for optimal and sustained (up to 6 h) nap-associated cognitive enhancements (Tucker et al., 2006). Thus, a longer nap encompassing both Stage 2 and SWS may be optimal.

Longer napping duration in this sample was also associated with higher levels of depressive symptomology. This is not surprising, as the authors' prior work has consistently shown strong relationships between depression and sleep–wake disturbance (McKinnon et al., 2014; Naismith et al., 2010, 2011a). Older adults with depression show higher levels of sleep and circadian disruption, including late insomnia (Naismith et al., 2011a). Thus, the effects of such sleep disturbance would undoubtedly contribute to daytime fatigue and sleepiness, and may increase the propensity for napping. Sleep–wake disturbance in older patients with depression has been shown to be related to neuropsychological functioning; however, the current findings were independent of the association between napping and cognitive functioning, as this was still significant even after controlling for depressive symptoms. Nevertheless, there is still the possibility that depressive, cognitive and sleep-disturbance features may share common neurobiological underpinnings, which may aid in earlier identification of disease onset. Additionally, antidepressant medication was associated with higher frequency of napping, which could relate to its tendency to have adverse effects on sleep architecture and capacity to induce daytime fatigue (Armitage, 2000).

The major limitation in this naturalistic study pertains to the measurement of napping. Cognitive outcome measurements were taken before the collection of napping characteristic measures, and not following specific napping episodes. Thus, any time-synched cognitive benefits that would be observable following a nap would not have been captured in this study. Similarly, detrimental or beneficial effects on sleep quality could be captured more precisely when analysed in relation to the subsequent nocturnal sleep period, as opposed to being averaged over 14 days. Finally, while standardised techniques using actigraphy to measure napping have been validated, the use of actigraphy to define napping is a methodology that requires further validation to confirm objectively that sleep is actually occurring.

It is also important to recognise that this sample was heterogeneous in the makeup of cognitive decline, as well as depression, history of depression and other variables. Nonetheless, this is quite reflective of the changes observed in older adults ‘at risk’ of dementia, and allows the relationship between these features and sleep characteristics to be examined in a broad sample of this population.

Napping may elicit benefits day-to-day based on the characteristic and level of demands placed upon the individual. In particular, there is a need to examine the benefits of planned and controlled napping, as opposed to merely analysing retrospectively the correlates of uncontrolled and unplanned napping. Short naps have been shown to improve alertness and cognitive processing in younger cohorts (Brooks and Lack, 2006; Tietzel and Lack, 2001). Planned and controlled napping may be a feasible and cost-effective method for improving declarative (e.g. episodic or semantic) memory in people with a disorder of memory consolidation, and specifically in those with amnestic MCI. Memory deficits are a prominent and defining feature of amnestic MCI that have pervasive impacts on a person's social and functional capacity. Further research is now warranted to determine whether planned and controlled napping is beneficial for cognitive processing in older adults ‘at risk’ of dementia.

Overall, this study has shown that napping is highly common in ‘at risk’ older adults; however, napping habits appear not to be consistent between or even within subjects. Even so, napping is significantly correlated with increased levels of depressive symptomatology and cognitive deficits, and this may be clinically relevant. While these findings are aligned with those observed in community-based studies of older people, they extend further by suggesting that napping characteristics should be more routinely monitored and investigated in clinical samples. In particular, in adults ‘at risk’ of dementia, napping warrants further attention from a research and clinical perspective.

Author contribution

NC contributed to analysis of results and drafting the manuscript; ZT contributed to study design and implementation, and critique of the manuscript; NR contributed to study design and critique of the manuscript; SD contributed to analysis of results, and drafting the manuscript; IH contributed to study design and implementation; SL contributed to study design and critique of the manuscript; SN contributed to study design and implementation, analysis of results and drafting the manuscript.