Intermediate, But Not Extended, Afternoon Naps May Preserve Cognition in Chinese Older Adults

Abstract

Background

The association between daytime napping and cognition is not well-studied in older adults. This study aimed to examine the associations between self-reported afternoon nap duration and change in cognition after 2-year follow-up in Chinese older adults.

Methods

A total of 3,037 individuals aged 60 years and older from the China Health and Retirement Longitudinal Study baseline and 2-year follow-up were included. Overall cognition was assessed by three interview-based sub-measures of orientation to time and attention, episodic memory, and visuospatial abilities. Change scores in cognition were calculated within subjects as follow-up minus baseline levels. Based on self-reported nap duration, older adults were grouped into: (i) Non-nappers (0 minutes); (ii) Short nappers (<30 minutes); (iii) Moderate nappers (30–90 minutes); and (iv) Extended nappers (>90 minutes) at baseline and follow-up.

Results

Change in cognition was significantly associated with both baseline napping and changes in napping habits from baseline to follow-up, controlling for age, gender, education, body mass index, depression, mobility, instrumental activities of daily living, nocturnal sleep duration, and cognition at baseline. Extended nappers at baseline performed significantly worse with respect to change of overall cognition from baseline to follow-up than non-nappers, short nappers, and moderate nappers. People who napped less than 90 minutes at both assessments performed significantly better from baseline to follow-up compared to non- or extended nappers at both assessments.

Conclusion

Results suggest that afternoon naps less than 90 minutes may be beneficial for cognition in older adults, while long afternoon naps may be detrimental.

More than half of Chinese older adults routinely nap during the day and consider napping to be beneficial for their health (1). The association between napping and cognition is complicated and not well studied in older adults (2,3), despite the fact that cognitive benefits of napping are well-established in younger adults (4). Daytime napping in older adults may compensate for insufficient nighttime sleep, prevent or reduce fatigue and daytime sleepiness, and restore energy, potentially enhancing cognitive performance (5,6). However, long naps may contribute to both sedentary lifestyle and impaired social engagement, both of which are associated with decreased cognitive performance (7). Therefore, it is important to identify the optimal napping duration for preserving cognitive function in older adults.

In addition to duration, the frequency, purpose, and circadian timing of naps are crucial to achieving cognitive benefits (6). Older adults may have frequent unintentional dozing during the day due to medications, chronic disease, and a lack of activity (8,9), which may lead to different cognitive effects than planned naps (10). Also, studies suggest that naps taken during the post-lunch dip period/afternoon (eg, 1 PM–4 PM) are more restorative than naps in the early morning or evening (6,11). Long naps and naps taken close to bedtime may affect sleep/wake circadian rhythm and disturb nocturnal sleep (6). Furthermore, the impact of sleep inertia following long naps on older adults’ cognitive impairment is unknown.

Current evidence suggests that longer naps (eg, >90–120 minutes) are associated with decreased cognitive performance or cognitive impairment in older adults (12–14). However, there is sparse and conflicting evidence related to the impact of shorter duration naps on cognition (15,16). Our prior study examined the cross-sectional association between afternoon nap duration and cognitive function in a nationally representative sample of Chinese older adults using the China Health and Retirement Longitudinal Study (CHARLS) baseline (13). Our findings suggested that routinely napping for 30–90 minutes (moderate napping) was associated with better overall cognition (a composite score of measures of orientation to time and attention, episodic memory, and visuospatial abilities) compared to no napping, less than 30-minutes napping (short), and more than 90-minutes (extended) napping.

To extend our previous work, we examined the associations of baseline nap duration and changes in napping habits from baseline to follow-up with changes in cognition after 2 years using the CHARLS baseline and 2-year follow-up datasets. We hypothesized that moderate nappers at baseline would maintain better cognitive performance over the follow-up period, and that groups that remained or became non- or extended nappers would have larger decreases in cognition over time compared to people that were moderate nappers at both baseline and follow-up.

Methods

Study Population

This was a retrospective study conducted with the CHARLS baseline (2011) and follow-up (2013) datasets. CHARLS is a nationally representative longitudinal study of Chinese community-dwelling residents 45 years or older intended to inform health and social economic changes related to China’s rapid aging and assist scientific research (17). CHARLS offers a wide range of information from sociodemographic factors to health (13,18).

A final sample of 3,037 respondents was included in analysis. Respondents were included in the current analysis if they: (i) were aged 60 or older; (ii) had complete data related to afternoon (post-lunch) napping duration and cognition measures at both waves; and (iii) had age, gender, education, body mass index (BMI) data at baseline.

Measures in CHARLS

Cognitive assessment

Cognition was assessed using three measures at baseline and follow-up that capture cognitive domains of mental intactness (orientation to time and attention), episodic memory, and visuospatial abilities, including: (i) the Telephone Interview of Cognitive Status (TICS-10); (ii) word-recall; and (iii) figure-drawing (13,19). Consistent with previous CHARLS publications, a measure representing respondents’ overall cognitive status that incorporates these three cognitive assessments was considered the primary measure of cognition; the three individual cognitive assessments are treated as secondary outcomes (13,19). The overall cognitive score is calculated as the sum score of TICS-10, word-recall, and figure-drawing, and ranges from 0 to 21. The three individual cognitive tests are summarized briefly below.

TICS-10 includes 10 items that measure the intactness of mental status (orientation to time and attention) (19). The TICS-10 score ranges from 0 to 10. The word-recall test assesses episodic memory and is the average number of correct immediate and delayed recalls from of a list of 10 Chinese nouns (18,19). The word-recall score ranges from 0 to 10. The figure-drawing test measures the respondent’s visuospatial abilities (18). In this task, respondents are shown two overlapped pentagons and asked to replicate the picture in a drawing. Respondents who successfully complete the task received a score of “1” and those who fail the task receive a score of “0.”

The overall cognition scores at both baseline and follow-up were computed for all 3,037 respondents. Change scores of overall cognition, TICS-10, and word-recall were calculated within subject as follow-up minus baseline levels. Positive change scores represent better performance from baseline to follow-up, while negative scores represent worse performance. Changes in figure-drawing (success/failure) were assessed using binary variables at baseline and follow-up (see Statistical Analysis).

Afternoon napping

Our two primary measures of afternoon napping were baseline napping groups and the change in napping habits from baseline to follow-up. Change in napping habits from baseline to follow-up represents the combination of baseline napping group and follow-up napping group and is henceforth referred to as “baseline/follow-up napping groups.”

Afternoon napping duration was assessed at both baseline and follow-up by asking “During the past month, how long did you take a nap after lunch in general?” Consistent with existing literature (1,13,15), respondents were categorized into four napping groups at both baseline and follow-up: non-nappers (0 minutes), short nappers (<30 minutes); moderate nappers (30–90 minutes); and extended nappers (>90 minutes).

We found no significant differences in change in overall cognition or change in any of the three sub-assessments of cognition between subjects who were short and moderate nappers at baseline; therefore, we combined short and moderate nappers into one group named intermediate nappers (ie, they took a nap, but it was less than 90 minutes in length) in analyses of changes in nap habits from baseline to follow-up. Thus, nine groups were created to capture the combination of baseline/follow-up napping habits: Non/Non-nappers; Non/Intermediate nappers; Non/Extended nappers; Intermediate/Non-nappers; Intermediate/Intermediate nappers; Intermediate/Extended nappers; Extended/Non-nappers; Extended/Intermediate nappers; and Extended/Extended nappers. The ultimate goal of these groupings was to assess the relationship between cognition and clinically interpretable changes in napping habits over time (eg, consistently taking intermediate duration naps or changing from no naps to extended naps), which may in turn help inform clinical guidelines.

Covariates

Baseline measures of demographics, health habits, self-reported comorbidity, functional status, social activities, nighttime sleep duration, and cognition measures were considered as covariates. Detailed descriptions of all study covariates have been provided previously (13) and are briefly described below.

Demographics included age, gender, education, and marital status. Education was grouped as lower than primary school and primary school or above. Marital status was grouped as married and non-married.

Health habits included respondents’ BMI, smoking (nonsmokers and smokers), and drinking habits (nondrinkers and drinkers).

Self-reported comorbidity was represented by three separate variables, including the total number of self-reported chronic diseases, the total number of self-reported medications for treating those diseases, and depression [measured by the 10-item Center for Epidemiologic Studies Depression Scale (CES-D 10) (20)].

Functional status was measured with instrumental activities of daily living (IADLs) and mobility status (21). These two measures were coded as dichotomous variables, where “0” represents no difficulty in IADLs/mobility and “1” represents having one or more difficulties in IADLs/mobility.

Social activities were assessed by asking respondents’ weekly participation in social related activities, such as interacting with friends or assisting others and separated into three categories: almost no social activity (less than once a week), some social activity (once a week), and socially active (more than once a week) (13).

Respondents were classified into four nighttime sleep groups based on self-reported nighttime sleep duration: very short sleeper (<5 hours/night), short sleeper (5 to <7 hours/night), normal sleeper (7 to <9 hours/night), and long sleeper (≥9 hours/night) (22).

Statistical Analysis

Differences among baseline napping groups were assessed using an analysis of variance (ANOVA) for normally distributed variables, a non-parametric Kruskal–Wallis test for skewed continuous data, and a chi-squared test for categorical data. Unless otherwise noted, continuous variables are summarized using means and standard deviations and categorical variables using percentages. We calculated the change scores for continuous cognition measures (overall cognition, TICS-10, and word-recall) as described above and found the change scores were normally distributed, thus allowing for parametric analysis on untransformed variables. In addition to examining the combined word-recall score, we performed sensitivity analyses assessing immediate and delayed word-recall, separately. Adjusted analyses of associations between baseline and baseline/follow-up napping groups and continuous change scores were performed using linear regression models. To examine the relationship between napping and change in figure-drawing success/failure, we utilized repeated measures logistic regression models accounting for repeated measures per individual, and assessed the significance of the time by napping group interaction term.

To comprehensively control for any possible confounding, we first fit a fully adjusted model including all relevant covariates: age, gender, education, marital status, BMI, depression, number of medications, number of self-reported chronic diseases, drinking, smoking, IADLs, mobility status, social activity, and nighttime sleep duration. Final models testing associations between napping and cognition included only those covariates found to be significantly associated (p < .05) with overall cognition in the fully adjusted models. Results from the final and fully adjusted models were similar. Therefore, only results from the final adjusted models are reported here. Given a single a priori primary outcome of change in overall cognition, statistical significance for primary associations with overall cognition was set at p < .05. We examined associations with changes in the three individual sub-measures of cognition, with significance based on a Bonferroni corrected p < .0167 (equals 0.05/3). Significance in all other analyses was based on a p < .05.

Results

Sample Characteristics

The sample was comprised of both male (53.8%) and female (46.2%) older adults with a low level of education (75.5% did not complete primary school). A total of 1,746 respondents (57.7%) reported taking an afternoon nap, with a mean duration of 62.7 (SD = 60.0) minutes at baseline, while 1,886 respondents (62.1%) reported taking an afternoon nap with a mean duration of 68.2 (SD = 60.0) at follow-up. A total of 1,334 respondents (43.9%) declined in overall cognition, with a mean decline of 2.67 points (SD = 1.97). For the sub-measures of cognition, 1,093 respondents (36.0%) declined in TICS-10 scores [mean decline of 2.19 points (SD = 1.45)], 1,335 (44.0%) respondents declined in word-recall [mean decline of 1.61 points (SD = 1.06)], and 493 (23.5%) of the 2,095 respondents who successfully completed figure-drawing at baseline failed the drawing test at follow-up. There were significant differences (p < .05) between baseline napping groups with respect to sex, education, smoking, drinking, BMI, depression, and nocturnal sleep duration. Detailed characteristics of the sample overall and by baseline napping groups are presented in Supplementary Table 1.

Associations Between Baseline Afternoon Napping Groups and Changes in Cognition

Baseline napping groups were significantly associated with change in overall cognition. Specifically, there were statistically significant differences among the napping groups for change in overall cognition (p = .002), TICS-10 (p = .008) and word-recall (p = .016) in final adjusted models.

Extended Nappers at Baseline Performed Worse Than the Other Three Groups

Overall cognition measure

When (see Table 1 and Supplementary Figure 1) examining the between group differences in change in overall cognition, extended nappers at baseline performed worse (lower change score) than the other three groups. Within baseline napping groups, there was a non-significant decrease in overall cognition at follow-up in extended nappers (p = .125), while both moderate (p < .001) and non-nappers (p = .012) had increased overall cognition at follow-up and short nappers showed a non-statistically significant increase (p = .076). The change in cognition in extended nappers was significantly different from changes in non-nappers (p = .011), short nappers (p = .019), and moderate nappers (p < .001). Differences in cognitive change scores between extended nappers and the three other baseline napping groups ranged from 0.41 to 0.63 points, which is similar to the differences in cognitive changes observed for a 7–10 year difference in baseline age within our sample [β (95% CI) for a 1-year change in age = −0.06 (−0.08, −0.04); p < .001]. While moderate nappers had the largest increase in overall cognition from baseline to follow-up on average, there were no other statistically significant differences among the baseline napping groups.

Table 1.

Change in Overall Cognition: Within and Between Group Comparison Among Baseline Afternoon Napping Groups

Estimates	Overall cognition
Within group estimates	Meana	95% CI	p †
Non-nappers	0.19	0.04, 0.34	.012
Short nappers	0.29	−0.03, 0.62	.076
Moderate nappers	0.41	0.25, 0.58	<.001
Extended nappers	−0.22	−0.49, 0.06	.125
Between group estimates	βb	95% CI	p ¶
Overall associations**			.002
Non vs. Short nappers	−0.10	−0.46, 0.26	.582
Non vs. Moderate nappers	−0.22	−0.44, 0.00	.054
Short vs. Moderate nappers	−0.12	−0.48, 0.24	.518
Non vs. Extended nappers	0.41	0.09, 0.73	.011
Short vs. Extended nappers	0.51	0.08, 0.94	.019
Moderate vs. Extended nappers	0.63	0.31, 0.95	<.001

Notes: Estimates are from the final parsimonious model adjusting for significant predictors (age, gender, education, BMI, depression, IADL, morbidity, and baseline cognition). Bold values represent p < .05.

^aModel estimated within group mean change.

^bEstimate of difference in change scores between indicated groups.

^† p-value testing whether within group mean change is different from 0 or within group odds ratio is different from 1; ^¶p-value testing whether there is a significant difference in changes between groups; ^**p-values for overall associations represent the probability of any significant differences in changes among the four baseline napping groups.

Individual cognitive sub-measures

We (see Supplementary Table 2) next examined differences in individual cognitive measures. Moderate nappers at baseline performed significantly better (higher change score) than extended nappers (p = .001) and non-nappers (p = .026) in change of TICS-10 (orientation to time and attention). Within each napping group, TICS-10 score in moderate (p < .001) and non-nappers (p = .001) significantly increased from baseline to follow-up, and showed a borderline non-significant increase in short nappers (p = .065). There was no change in TICS-10 within the extended nappers (p = .816).

For changes in word-recall (episodic memory), extended nappers did worse (lower change score) than non- (p = .007), short (p = .008) and moderate nappers (p = .003). Results were similar when examining immediate or delayed word-recall separately (Supplementary Table 3). When examining within group changes, word recall decreased significantly in extended nappers from baseline to follow-up (p = .013), while there were no significant changes within the other three napping groups.

Significant higher proportion of baseline non-nappers failed in the figure-drawing test (visuospatial ability) at follow-up than baseline. No significant findings were observed in between group comparisons and other within group comparisons of change in figure-drawing from baseline to follow-up.

Associations Between Baseline/Follow-up Napping Groups and Change in Cognition

Baseline/follow-up napping groups were associated with changes in cognition. Specifically, baseline/follow-up napping groups were independently associated with change scores of overall cognition (p < .001) in final adjusted models. The associations with change scores in TICS-10 (p = .021) and word-recall (p = .030) were nominally significant, but did not meet the Bonferroni adjusted p-value (p < .0167).

Intermediate/Intermediate Nappers Performed Better Than Non/Non, Extended/Extended, and Extended/Non-Nappers

Overall cognition measure

For the (see Table 2 and Supplemental Figure 2) between group comparisons, we compared changes within each baseline/follow-up napping group to those among older adults who were intermediate nappers at both baseline and follow-up, as this was the group with the greatest increase in cognition. Among those with consistent napping patterns, older adults who were non-nappers (non/non) and extended nappers (extended/extended) performed worse from baseline to follow-up (lower change score) than intermediate/intermediate nappers (p = .014; p < .001). Among groups that changed napping habits from baseline, only older adults that switched from extended to non-nappers performed significantly worse from baseline to follow-up (lower change score) than intermediate/intermediate nappers (p < .001). Other groups (extended/non; non/intermediate; non/extended; intermediate/non; intermediate/long; long/intermediate) were not significantly different from the intermediate/intermediate group, although all had less positive changes.

Table 2.

Change in Overall Cognition: Within and Between Group Comparison Among Baseline/Follow-Up Afternoon Napping Groups

Estimates	Overall cognition
Within group estimates	Meana	95% CI	p †
Intermediate/Intermediate	0.51	0.33, 0.69	<.001
Non/Non	0.18	−0.01, 0.37	.066
Non/Intermediate	0.28	0.00, 0.57	.051
Non/Extended	−0.02	−0.54, 0.50	.940
Intermediate/Non	0.21	−0.12, 0.53	.211
Intermediate/Extended	0.16	−0.19, 0.52	.364
Extended/Non	−0.98	−1.75, -0.21	.012
Extended/Intermediate	0.10	−0.31, 0.51	.622
Extended/Extended	−0.33	−0.75, 0.10	.133
Between group estimates	βb	95% CI	p ¶
Overall associations**			<.001
Intermediate/Intermediate: Reference group
Non/Non	−0.33	−0.60, −0.07	.014
Non/Intermediate	−0.23	−0.57, 0.11	.192
Non/Extended	−0.53	−1.08, 0.02	.061
Intermediate/Non	−0.30	−0.68, 0.07	.113
Intermediate/Extended	−0.35	−0.74, 0.05	.088
Extended/Non	−1.49	−2.28, −0.70	<.001
Extended/Intermediate	−0.41	−0.85, 0.04	.075
Extended/Extended	−0.84	−1.30, −0.37	<.001

Notes: Estimates are from the final parsimonious model adjusting for significant predictors (age, gender, education, BMI, depression, IADL, morbidity, and baseline cognition). Bold values represent p < .05.

^aModel estimated within group mean change.

^bEstimate of difference in change scores between indicated groups and the intermediate/intermediate group.

^† p-value testing whether group specific mean change is different from 0 or odds ratio is different from 1; ^¶p-value testing whether there is a significant difference in changes between groups; ^**p-values for overall associations represent the probability of any significant differences in changes among the nine temporal napping groups.

When examining the within group change, intermediate/intermediate nappers had a significant increase in overall cognition (p < .001) from baseline to follow-up, while non/non-nappers had a borderline non-significant increase (p = .066) and extended/extended nappers had a non-significant average decline (p = .133). The increase of overall cognition among intermediate/intermediate nappers was 0.84 points greater (better) than the change in extended/extended nappers and 0.33 points greater than the change seen within non/non-nappers. On the other hand, older adults that switched from extended napping at baseline to non-napping at follow-up had the largest decrease in overall cognition [mean (95% CI) = −0.98 (−1.75, −0.21); p = .012]. This corresponds to about a 1.5 point difference in change scores between intermediate/intermediate and extended/non-nappers within our sample (p < .001). Given our observation of an expected 0.06 point difference in cognitive change for each 1 year difference in age, these effects are similar to those expected for anywhere from a 5 to 25 year difference in baseline age.

Individual cognitive sub-measures

The (see Supplementary Table 4) change score of TICS-10 (orientation to time and attention) from baseline to follow-up in the intermediate/intermediate group was significantly better (higher change score) than changes seen within non/non (p = .039), extended/extended (p = .007), and extended/non (p = .001) napping groups. Intermediate/intermediate nappers had the largest increase in TICS-10 [mean (95% CI) TICS-10 change = 0.38 (0.25, 0.51); p < .001] from baseline to follow-up. TICS-10 in Non/non (p = .008) and Intermediate/non-nappers (p = .029) also increased from baseline to follow-up.

For word-recall, intermediate/intermediate nappers had a significantly higher change score than extended/extended nappers (p = .001). Results were similar when examining immediate or delayed word-recall separately (Supplementary Table 3). When examining the within group change, extended/extended nappers were the only group that experienced a significant decrease in word-recall scores (p = .024) and intermediate/intermediate nappers were the only group had a significant increase in word recall scores (p = .005) at follow up.

Significant higher proportion of non/non-nappers failed in the figure-drawing test (visuospatial ability) at follow-up than baseline. No significant findings were observed in between group comparisons and other within group comparisons of change in figure-drawing from baseline to follow-up.

Discussion

We examined the associations between self-reported afternoon napping duration and changes in cognition measures at a 2-year follow-up within a nationally representative sample of 3,037 community-dwelling Chinese older adults. Our results indicate that intermediate napping may be most beneficial for preserving cognition over 2 years, while extended napping is associated with more cognitive decline on average. These findings support and extend those from our prior cross-sectional study within the CHARLS national baseline data, where we found an inverted U-shape association between the baseline napping groups and overall cognition; absence of napping or napping for more than 90 minutes were associated with worse cognition and naps of a moderate duration were associated with better cognition (13).

Within our sample in the present manuscript, overall cognition was significantly increased at the 2-year follow-up in non- and moderate- nappers at baseline and intermediate nappers at both assessments. It was significantly decreased at follow-up in extended/non-nappers. Extended nappers at baseline performed significantly worse (lower change score) in change in overall cognition when compared with all other baseline napping groups. In addition, compared with intermediate nappers at both assessments, those who reported no naps or extended naps at both baseline and follow-up, and those who switched napping habits from extended napping to no napping, had significantly lower change scores (performed worse) in overall cognition.

The protective effects of intermediate napping on cognition have also been reported in prior research. Campbell and colleagues found that afternoon napping (mean [SD] duration of 81 [25.9] minutes) was associated with significant improvements in cognitive and psychomotor performance on both the same day and the day following napping in 32 older adults (16). A month-long napping regimen (sleep opportunity in the afternoon) has also been shown to improve cognitive performance in 22 older adults (3). Findings from a study of 337 patients with a diagnosis of probable Alzheimer’s disease (AD) showed that napping for less than 60 minutes protected against the development of AD in 5 to 10 years (23). However, one study found no effects of a 17-day long, 90-minute afternoon nap regimen on cognitive performance in a small sample of nine relative healthy older adults (15).

Intentional daytime napping limited to a certain duration could provide older adults with cognitive benefits by increasing 24-hour total sleep time (3) and avoiding or reducing daytime sleepiness (6). However, unintentional dozing or uncontrollable napping may be a sign of disease (eg, dementia), rather than a good napping practice for better health (10). In addition, napping in the afternoon (between 1 PM and 4 PM) may produce more cognitive benefits than naps taken in other times of the day without interrupting nighttime sleep (3,16). According to the specific nap question asked (“During the past month, how long did you take a nap after lunch in general?”), we assume the self-reported naps in this study were planned/intentional naps that are taken routinely during the post-lunch dip period. However, it is possible that naps reported in our sample might not have been planned, but rather have resulted from other conditions, such as sleep disordered breathing and excessive daytime sleepiness. We were unable to adjust for these possible confounders in our analysis due to absence of measures in CHARLS.

The increased cognition score at the follow-up found in some napping groups in our sample may seem inconsistent with the current evidence on age-related decline in cognitive function in older adults (24). In addition to the cognitive benefits from certain nap habits discussed above, other factors may have contributed to these apparent increases in cognition measures. For example, the majority of our sample (74.3%) was younger aged older adults (60–69 years), who may not be expected to show as much cognitive decline as more older aged adults (>70 years). Also, the same cognitive assessments were applied at both baseline and the 2-year follow-up. Thus, improvements observed at the follow-up assessment could, in part, be attributed to practice effects (25). Since CHARLS did not record whether participants recalled specific questions from their baseline assessment, we were unable to control for these effects; future research should examine the relative impact of practice effects compared to benefits of behavioral interventions such as napping.

The suggested detrimental effects of extended naps with cognition are consistent with prior research. One cross-sectional study with 2,932 elderly women reported that napping for a duration of two or more hours per day was associated with a higher risk of cognitive impairment (12). A more recent cross-sectional study in 133 older adults “at risk of dementia” reported that older adults who napped for longer tended to have poorer cognitive performance in terms of processing speed, verbal fluency, and verbal memory (17).

Of note, people who changed napping habits from extended to no napping had the largest declines in cognition compared to all other baseline/follow-up napping groups. This change in napping habits could be caused by medical conditions, such as pain (26), which is known to increase older adults’ risk of cognitive decline (27). As only 49 respondents (12.9% of baseline extended nappers) were in this group, this association should be examined in larger samples of extended/non-nappers. Taken together, these findings suggest that napping less than 90 minutes is superior to non- and extended naps for preserving cognitive function over time.

To our knowledge, this study was the first to examine the association between daytime napping duration and each sub-measure of cognition. In our sample, baseline moderate nappers performed significantly better than both extended and non-nappers in terms of change in TICS-10 (orientation to time and attention). Baseline extended nappers performed worse in terms of change in word-recall (episodic memory) from baseline to follow-up than the other three baseline napping groups. We saw nominally significant associations of baseline/follow-up napping groups with change in TICS-10 and word-recall, although results were not statistically significant after Bonferroni adjustment. Compared with intermediate nappers at both assessments, all other baseline/follow-up napping groups had lower change scores in TICS-10 and word-recall, although not all differences were statistically significant. No association was found between napping groups and change in visuospatial ability (figure-drawing).

This study has some limitations. First, the self-reported measures of afternoon napping/sleep may introduce recall bias. Also, our analysis only included self-reported napping during the post-lunch dip period. We do not have information on daily napping frequency and napping duration at other times. Future studies should consider adding an objective sleep measure such as actigraphy to more accurately assess napping/sleep parameters, as well as collect data on the timing, frequency, duration, and intention of the nap. Second, the capability of our overall cognition score to assess visuospatial ability is limited, since figure-drawing only contributes 0 or 1 point to the overall cognition score. However, the scoring method is similar to that used in the Montreal Cognitive Assessment (28) and the Mini-Mental Status Exam (29). In addition, although measurements were taken 2 years apart, the use of the same cognitive assessments at both the baseline and follow-up could have resulted in small increases in cognitive score due to practice effects (25). We could not examine the possible impact of remembering specific questions with the available data in CHARLS. Furthermore, we were unable to adjust for some prevalent conditions (eg, sleep disordered breathing and excessive daytime sleepiness) in older adults that may contribute to extended napping and cognition levels (30,31) due to absence of these measures in CHARLS. Moreover, our findings from a population of Chinese older adults may not be generalizable to other ethnic groups. Finally, our analyses only included one follow-up period, 2 years after baseline. Multiple wave data within a longer follow-up duration may be more informative to describe the impact of napping on the trajectory of cognitive change.

Despite these limitations, this study has several strengths. First, we conducted longitudinal analyses within a nationally representative cohort of Chinese older adults. Second, we categorized napping into four groups, which allowed us to examine the associations of napping and cognition in a clinically relevant manner and more specifically than studies that have only two napping groups (10,12). Most importantly, to the best of our knowledge, this study is the first study that examined the relationship between changes in napping habits over a 2-year period and changes in cognition in older adults, which provides additional evidence on associations of napping and cognition.

In summary, this study has shown that napping for a duration of more than 90 minutes may be detrimental to cognition, while a nap limited to at most 90 minutes may be beneficial for preserving or improving cognitive function in older adults over 60 years of age. These findings will help inform clinicians when providing older adults with good daytime sleep recommendations in order to optimize cognition. Napping is a low cost intervention that is easily scalable to large populations as a means to potentially protect cognitive function. Prospective experimental studies are needed with objective assessments of sleep and napping to further examine the impacts of daytime napping on cognition in older adults, as well as the efficacy of specific interventions with napping protocols.

Supplementary Material

Supplementary data is available at The Journals of Gerontology, Series A: Biological Sciences and Medical Sciences online.

Funding

This work was supported by the National Institutes of Health T32 HL07713.

Conflict of Interest

Nothing to report.